The Bloedel Reserve Report

Page 1

The Bloedel Reserve Report

A Data Discovery & Analysis Experiment K. W. Bridges 2019 1


2019 Š K. W. Bridges www.kimbridges.com kim.bridges@gmail.com


Preface I’m curious how the Bloedel Reserve fits into America’s community of botanical gardens. I also wonder how visitors perceive this garden. Those concerns started an experiment. The overarching question: How much can I learn about the Bloedel Reserve from harvesting opensource resources? This wasn’t a neat experiment. It was more a bumpy trip through a forest of data and the results of battles with truculant analytical machinery. Like many research projects, the more I discovered, the more subjects I found awaiting analysis. The deeper I went into an analysis, the broader the base became. My conclusion, after seeing the results of my analyses, is that there is a rich trove of data in our open-source resources. Many institutions, not just botanical gardens, are likely to benefit from harvesting the data and subjecting the numbers to analysis. I hope that this report aids others in this task.

1


2


Bloedel Reserve: A Small Introduction TL;DR. The Bloedel Reserve is a special place, created with the vision, resources and personal effort of Prentice Bloedel. It took many years to rehabilitate a damaged landscape. This effort involved many people. After years of private development, the Reserve is now open to the public. Visitors wander through a series of contrasting landscapes, some are natural and others carefully constructed with a respect for nature. It’s likely your visit will “inspire and refresh,” and, thereby, fulfill Mr. Bloedel’s goal in making the Reserve a public resource.

Before it became the Bloedel Reserve, this property at Agate Point on Bainbridge Island passed through many hands. This was an area of forest entrepreneurship. Nearby mills cut trees logged from the Bainbridge Island woodlands. Fires further impacted the area. People bought, sold and subdivided the land. This was not a pristine wilderness. A large purchase, in 1904, began the process of property consolidation. This started the Collins Estate. At first, life here was simple. A rustic seashore cabin was adequate. Eventually, Mrs Collins built a fashionable estate, called Collinswood. They completed the elegant residence in 1932. They added a small garden and planted some trees and shrubs around the property. This was the Collins family summer home. Collinswood caught the attention of Mr. and Mrs. Bloedel when it came up for sale. They purchased the property in 1951. It would become their permanent residence for many years. Prentice Bloedel was a forestry company executive. It was a lucrative occupation. By the time he retired, he was a wealthy man. Kreisman (1988) has provided key access to understanding how the Re3


serve developed. His book includes a revealing statement by Prentice Bloedel. These words1 provide a starting point for understanding the Reserve and help define its goals. With the exception of a tiny garden around the house the whole property was covered with a young mixed conifer hardwood stand, two streams, swamps and brushy patches typical of so much cutover forest land in the Pacific Northwest. In spite of our citified pasts it wasn’t long before we were exploring the old logging roads that threaded the woods, heading out cross country to see what we could find, following fence lines. In the course of these forays we found the land itself marvellously varied in contour and physiography. We found single plants and colonies of fragile woodland species, mosses, ferns, a world of incomparable diversity, a panorama of survival in an eternal struggle, exciting in its vitality. We found that plants often have a way of arranging and disposing themselves with a harmony of color, texture and form when left to themselves. We discovered that there is grandeur in decay; the rotten log hosting seedlings of hemlocks, cedars, huckleberries, the shape of a crumbling snag. Out of these experiences comes an unexpected insight. Respect for trees and plants replaces indifference; one feels the existence of a divine order. Man is not set apart from the rest of nature -- he is just a member of that incredibly diverse population of the universe, a member that nature can do without but who cannot do without nature. One realizes that we humans are trustees in this world, that our power should be exercised in this context. The new awareness determined us to set the land aside for the primary purpose of providing others with the opportunity to enjoy plants both as arranged by man 4


and as they arrange themselves; and for the purpose of providing people wandering about the Reserve a refreshing experience of nature and a broadening of their appreciation of their world. Prentice Bloedel’s goal was to transform a cut-over, second-growth forest into a “an example of man working harmoniously with nature, where his power to manage is used cautiously and wisely.”2 Further, he felt that with “study, understanding and sympathetic treatment, the whole property would possess an internal unity and integrity that would realize its capacity to inspire and refresh.”3 The lofty goal meant that a lot of hard work would go into the project. A series of noted landscape architects help fulfill Bloedel’s vision. Thomas Church was the primary designer of the Reserve in the early days. Fujitaro Kubota created the Japanese Garden. Many great botanical gardens begin as a project related to a wealthy family. This is true of the Bloedel Reserve, as well as other notable gardens like Huntington Gardens and Library and Longwood Gardens. There comes a time of transition, when the founder transfers ownership and maintenance to another entity. This new owner is often a non-profit foundation or government entity. Mr. Bloedel transferred ownership of the Reserve to the University of Washington in 1970. This new relationship was to assure that the public would benefit from the Reserve. A non-profit foundation, the Arbor Fund, was created to provide for the garden’s maintenance. The Bloedels endowed this foundation so it could function in this role. The University relationship did not last. In 1985, the Arbor Fund purchased the Reserve. The Arbor Fund assumed the role of developing the Reserve. In particular, they needed to resolve the use of the Reserve by the public. The development of the Reserve was not complete. Although the Arbor Fund was in charge, Prentice Bloedel remained active in all aspects of the garden’s activities. 5


The Arbor Fund held a design competition in 1978. The goal was to select a landscape architect with a plan that would carry out the grander visions of the Bloedels and the Arbor Fund. Richard Haag’s plan won. He worked closely with Mr. Bloedel for seven years. This was a fruitful collaboration. Mr. Haag was the second of the Bloedel Reserve’s great landscape architects. There were a lot of projects. Some work involved massive movements of soil. Some changes were subtle. Other activities transformed large areas. Richard Haag’s plans envisioned a series of gardens. His idea was that visitors would progress through a contrasting series of gardens. One garden might be formal. Another is in decay. You walk through natural environments, too. Today’s visitors experience this theme of contrasts. You stroll between places in the Bloedel Reserve, taking your time. There are places to stop and enjoy how nature is presented. Your experience changes as you move along the trail. The garden may even surprise you when an abrupt transition jolts your senses. Bloedel Reserve continues to evolve. Planning was once a years-long stroll with Mr. Bloedel and his landscape architect. It is now more formal. Leading design and planning firms create multi-year master plans. The Bloedel vision remains. The big change is the public now has access. Behind the scenes, work continues with what is likely the hardest part of maintaining Bloedel Reserve. The task: ensuring that there is sufficient funding. Even large endowments, once thought to be adequate, can wither with time. The public has access to the Bloedel Reserve. It has also gained the responsibility to contribute funds so that the Reserve will continue to “inspire and refresh” future generations. Prentice Bloedel died in 1996. He was 95 years old. 6


7


8


Objectives and Perspectives This report has several objectives. I refer to an “attraction” here as it is TripAdvisor’s category for places like botanical gardens. I’m using the term both in the specific instance of a botanical garden and more generally as a similar, public/private establishment that caters to and depends on visitors. • Establish if open-source data, such as Wikipedia and TripAdvisor, provide interesting and useful information about an attraction. • Create general analysis protocols for analyzing and visualizing opensource data that pertain to attractions. These objectives are addressed in a series of perspectives. • Fitting Bloedel Reserve into the Botanical Gardens • Visitors Impressions of Bloedel Reserve • Bloedel Reserve as a World-Class Botanical Garden There is, at the end, a discussion of data sources. Important Disclaimers: This is not a report about what the Bloedel Reserve should be doing. That is well beyond my intent. I also stop short, in most cases, in presenting detailed analyses of specific data. My emphasis is to show what data are available and suggest useful ways to manipulate and present this information.

9


Part I: Fitting Bloedel Reserve


e into US Botanical Gardens


Figure 1. Locations of the botanical gardens listed in Wikipedia for the lower-48 states.

Figure 2. Locations of the botanical gardens listed in Wikipedia for Hawai`i.

12


Botanical Gardens in the US Bloedel Reserve has a lot of company. The US, as a whole, has a lot of botanical gardens. A Garden Table1, derived mainly from Wikipedia lists, was created for these analyses. This dataset has 768 botanical gardens. You can see the distribution of the gardens across the US in two maps. Figure 1 shows where gardens are located in the US “lower 48.” The Hawai`i locations appear in Figure 2. Hawai`i is shown separately since using the same scale as the continental area would make the islands very tiny and completely obscure the garden distribution. The plotting symbols have some “transparency” so that multiple gardens in one location appear as darker symbols. The greatest abundance of continental gardens is in the North East. Florida is well represented, as are the West Coast cities in Southern California, the Bay Area, Portland and Seattle. Hawai`i has botanical gardens on four of the main islands. Alaska (not shown here) has three gardens. One each in Anchorage, Fairbanks and Juneau.

1 Data were compiled into tables, often using multiple sources. The details are described in Part IV.

13


Comparing the Gardens in Washington to other States We can see the number of botanical gardens in each state by using a bar chart (Figure 3). Here, the state of Washington is seen in the top dozen states in terms of the number of botanical gardens. This state has nearly twice the median number of botanical gardens. Using this ranking, the Washington state looks well supplied with botanical gardens. Adjusting for Population It is always a good idea to be alert to basic differences when making comparisons. In this case, states have significantly different populations. It is this population that usually funds and uses the gardens. States with more people would be expected to have more gardens. Washington’s population in 2017 was 7,405,743. This makes the state the 13th largest in the country. The simple adjustment is to normalize the data for population size. In this case, the number of gardens is divided by the state’s population in millions of people. If we express the number of botanical gardens on a per capita basis (Figure 4), Washington drops a bit to a tie with Connecticut for 17th place. Overall, Washington does quite well in terms of the number of botanical gardens. It might be noted that Hawai`i has a lot of gardens, but not a very large population. This propels that state to the head of the list of gardens per million residents.

14


Median number of gardens: 12

State

WY RI AK SD MS WV NH ID AR NM DE NV ND MT UT MO DC MD CO AZ ME KY KS CT MN MI LA WI NE IN SC OK IA GA VA MA OR WA TX NJ OH TN NC NY IL HI FL PA CA

0

20

40

Number of Gardens

60

Figure 3. The number of botanical gardens in each state based on the Wikipedia listing.

15


WV RI MA UT NM ID WA CT SC PA

State

NH KS TN AK SD OK OR IA DE MT ME ND DC HI 0

5

10

15

Gardens per Million Residents

20

Figure 4. The number of botanical gardens in each state per million residents, based on the Wikipedia listing. Only the states with the highest per capita number of gardens are listed.

16


17


Gardens are Abundant in Washington The Wikipedia list of US botanical gardens in the state of Washington has 23 entries. A quick web search shows there are several prominent gardens missing from this list (the Dunn, Seattle Japanese and Kubota gardens); these three gardens were added for the analyses here. There are likely quite a few more small or specialized gardens that are not in the state list. It is not the purpose of this study to be thorough in this regard. This is mostly a reminder that the Wikipedia list is incomplete in detail. Correcting the deficiencies is a difficult task. The approach used here, knowing that the list is incomplete, is to assume that the types of listing used here are adequate representations when the subject is viewed broadly. The main Garden Table was not modified for the purpose of mapping the locations of gardens in Washington. Instead, a new table was built, starting with each of the Washington state gardens entered into the table. The area of each garden was found, usually on a garden’s website. The location of each garden, as geographic coordinates, was extracted from a Wikipedia page describing the garden. The location often needed transformation so that all the coordinates were expressed in the same decimal format. The spread of the location marks (Figure 5) shows most gardens concentrated near Seattle. A second map (Figure 6) restricts the map’s extent to the Seattle area. This focus gives a better indication of the gardens near Bloedel Reserve. The Seattle region appears to be well provided with botanical gardens of a variety of sizes. For reference, Bloedel Reserve is one of the larger gardens with an area of 150 acres.

18


Garden Size (Acres) 50 100 150 200

Figure 5. The locations and relative sizes of botanical gardens in Washington from the Wikipedia listing.

Garden Size (Acres) 50 100 150 200

Figure 6. The locations and relative sizes of botanical gardens near Seattle, mostly from the Wikipedia listing.

19


What do you Call a “Botanical Garden”? The Bloedel Reserve is the focus of this study. The term “Reserve” seems quite unusual for a botanical garden. This prompts the question of what terms the US botanical gardens use to describe themselves. We turn again to the Wikipedia listing of US botanical gardens. A word cloud is constructed by analyzing the frequency of words in a body of text. In our case, the data come from the names of all 768 botanical gardens. Some preprocessing went into constructing the list of words. A set of common words (e.g., the, an, with) were removed from the list. Then two words that are either singular or plural were combined; gardens became garden and arboreta became arboretum. Two other terms were simplified; botanic became botanical and horticultural became horticulture. Then, a number of place name words (e.g., city, county, state), institutional designations (e.g., university, college), and geographic features (e.g., mount) were removed from the list. This wasn’t an exhaustive process but mostly one to basically clean up the data. The resulting list was analyzed for the frequencies of each word. Low frequency words were dropped. The words were then arranged into a diagram, with the size and color of each word showing its frequency. This diagram is a word cloud. The botanical garden word cloud (Figure 7) shows the most frequently used words in the names of the US gardens. Three terms (garden, botanical, and arboretum) totally dominate the graphic. The other terms are too small to be seen. That’s not very useful. The scaling of data is quite important and this procedure will be used quite often. 20


forest

botanical old

foundation

conservatory

center woods

acres

park

ranch

public

science

society

coast

heritage

western

zoological

research

native

reserve

sanctuary

garden cactus

international lake

japanese

plantation

cemetery horticulture

farm

area

plant spring

place

memorial nature

museum

lilac

zoo

cherokee

mansion

springs

arboretum hill

bird

art

american

wildflower

education

valley greenhouse

rose

tropical

preserve

trail

campus rhododendron

coastal

river

desert

riverside

Figure 7. Words used in the names of US botanical gardens. Note the use of synonyms in the discussion.

By using the log of the word frequencies (Figure 8), the other terms are seen more easily and you get a better balanced view of the names used for the garden names. Garden, botanical and arboretum are still obvious common terms. They are joined by park, center, museum, memorial and conservatory. Browsing the word cloud shows some of the diversity in the type of botanical garden. This might help explain why there are so many gardens in the US. The list of botanical gardens shows only three with “reserve” in the name: Bloedel Reserve, Shaw Nature Reserve in Dark Summit, MO, and Schmeekle Reserve in Stevens Point, WI. Being called a “botanical garden” or “arboretum” would imply the need to label the plants. This identification process, generally with a plaque with the scientific and family names, along with the common name and geographic origin of the species, is a standard applied to all “scientific” gardens. 21


It can be claimed that the identification signs, while useful to some people, actually detract from the visitor experience. Signs call attention to the trees, not the forest. The Bloedel Reserve experience, if a reading of the words of the garden’s architects is correct, is that the forest (sensu lato) is the important thing. If you’re interested in a name, you can look it up in a database. The Bloedel Reserve is a series of holistic creations. These are Landscape Architect Haag’s “rooms.” It’s why you move from place to place with each transition giving you a different (often contrasting) visual and emotional experience. Would the use of plant identification signs be disruptive? The name of the botanical gardens is important. It is, most likely, the first hint that at the Bloedel Reserve, you’re not visiting a traditional botanical garden or arboretum.

education

trail

springs wildflower sanctuary

international

river rhododendron desert cemetery science cactus horticulture foundation japanese nature place area art center mansion museum public american lake rose western forest bird acres preserve

woods

arboretum

reserve

valley

zoological

garden

botanical park hill plant

riverside conservatory research memorial coastal cherokee zoo campus farm tropical plantation greenhouse

coast

lilac native

heritage

old ranch spring

society

Figure 8. Frequencies of words used in names of botanical gardens. Frequencies were converted to log values.

22


23


Figure 9. USDA Plant Hardiness Zone map.

Figure 10. Legend to the USDA Plant Hardiness Zone map.

24


Botanical Gardens & Hardiness Zones The USDA Hardiness Zone system (Figure 9) gives a general indication of what plants will grow in a particular area. uite a few botanical gardens are found in most of the zones. The Bloedel Reserve is one of the country’s most northerly sites (excluding Alaska). Yet the Hardiness Zone is 8b, a relatively warm zone. That’s something a visitor from outside the state of Washington probably won’t expect. The Hardiness range for trees gives an idea of where they can grow. Some of the trees at Bloedel Reserve are shown in Figure 11. Nearly half appear to be growing outside their usual range.

Species

Western Red Cedar Western Hemlock Red Alder Quaking Aspen Portugal Laural Mountain Hemlock Madrone Korean Fir Katsura Grand Fir English Elm Empress Tree Douglas Fir Dawson Magnolia Cascara Bigleaf Magnolia Atlas Cedar

colder 1

3

5

7

USDA Hardiness Zone

9

Figure 11. The USDA Hardiness Zone range for some tree species found at Bloedel Reserve. The red line is the Zone for Bloedel Reserve.

25


2a 3b

Colder

4a 4b 5a

USDA Hardiness Zone

5b 6a 6b 7a 7b 8a Bloedel Reserve

8b 9a 9b

10a 10b 11a 11b

Warmer

12a 12b 0

25

50

75

Number of Botanical Gardens

100

Figure 12. The number of US botanical gardens in each of the USDA Plant Hardiness Zones. Bloedel Reserve is in the relatively warm zone 8b.

Botanical gardens have often been the site of new species introductions to a region. The plants are grown in the garden to see if they are adapted to the new environment. Most of the botanical gardens in the US are in colder Hardiness Zones (Figure 12). 26


Bainbridge Island Compared to Japan Japan is the origin of many non-native plants in the Bloedel Reserve. The USDA Plant Hardiness Zones similar to the one covering the Reserve are shown for Japan in Figure 13. The climatic parallels are obvious. Many plants native to Japan should grow well in the Bainbridge Island climate.

Figure 13. USDA Plant Hardiness Zones for Japan. Only zones 8a, 8b and 9a are shown.

27


28


Botanical Garden Sizes Larger gardens often have the room to offer more varieties of experiences to the visitor. They can also hold more plants (perhaps increasing the diversity). The Bloedel Reserve is on the larger end of the size scale at 150 acres, about 60 of which are developed (Figure 14). Note that the size graph uses a log scale for the sizes. This means that there are a lot of small gardens and a few really large ones.

60 â—?

Median area: 36.5 acres Bloedel Reserve

Number of Gardens

â—?

40

20

0 1

10

100

Garden Area (acres)

1000

10000

Figure 14. Sizes of US botanical gardens listed in Wikipedia for which data could be found. Note the use of a log scale for the areas.

29


Part II: Visitor Impressio


ons of Bloedel Reserve


32


Bloedel Reserve Rankings The many highly favorable TripAdvisor reviews of Bloedel Reserve have placed it as the top attraction on Bainbridge Island. That’s a notable achievement. TripAdvisor reviewers have been steady in providing top ratings and reviews. This provides a rich source of information to potential visitors and garden administrators. Additions to the TripAdvisor database add to the depth of discovery. Summary data, such as the top rank in “things to do in Bainbridge Island” is important, as this may be the fact that helps a visitor make a decision. “Let’s go to the top place!” is a motivating statement. Trends are equally important, especially to people responsible for operating an attraction such as Bloedel Reserve. “Are we getting fewer or more low ratings?” is a typical question. The answer might be important. Glancing through the TripAdvisor website provides some insight. Here, by extracting the data and visualizing it in unique ways, it is hoped that additional perspectives will be seen.

33


Reviewer Ratings Overall Ratings The “big picture,” besides being listed as number one, is the distribution of reviewer ratings. The TripAdvisor rating scale goes from 1 to 5. This system uses both numbers and a matched word scale. The words are “terrible,” “poor,” “average,” “very good,” and “excellent.” (The numbers and names will be used interchangably here.) No review has given a “terrible” rating to the Bloedel Reserve, so you’ll not see that term in the context of the Bloedel Reserve. Overall, the distribution of ratings (Figure 15) is very favorable. 81.2% of the ratings fall in the “excellent” category, the highest score (5 on the TripAdvisor scale). There are virtually no ratings in the “poor” and “average” categories. TripAdvisor implies that the average rating in their system is “average” (value 3). Clearly, Bloedel Reserve is seen as special according to the 490 reviews used in this analysis. Annual Rating Trend The annua trend in ratings (Figure 16) shows an increase over the 20122018 period analyzed. The slight drop off in 2018 is probably not a great cause for concern. TripAdvisor does say that a high comparative rating for the attraction in an area (which is currently 1 of 30 for Bloedel Reserve) depends not just on the rating value, but also the number of ratings and their currency. This is an overall trend that should be watched as a substantial decrease in reviews could lower the overall rating (i.e., Bloedel Reserve could slip from being the number one attraction on Bainbridge Island).

34


81.2 %

Note: All countries Total Reviews: 490

Number of Reviews

400

300

200

15.9 %

100

1.4 %

1.4 %

Poor

Average

0 Very Good

Rating

Excellent

Figure 15. The number of each rating for Bloedel Reserve over the period from 20122018.

Visitor Reviews

60

Note: Only USA

40

20

0 2012

2013

2014

2015

2016

Year of Review

2017

2018

Figure 16. The number of reviews each year for Bloedel Reserve.

35


Ratings Within the Annual Trend You can look deeper into the annual trends by displaying the number of each category of ratings for each year (Figure 17). The goal is to see if there are any significant changes in the rating distributions. There appears to be a consistent trend with mostly Excellent (5) and Very Good (4) ratings, year after year. The few Average (3) and Poor (2) are worth examining in detail by looking at the review comments. There are so few, however, that in the absence of other data, these low-rating reviews aren’t troublesome.

80 Note: All reviewers

Number of Ratings

60 Rating 2 3

40

4 5

20

0 2012

2013

2014

2015

Year

2016

2017

2018

Figure 17. The number of each rating for Bloedel Reserve, by year.

36


Reviewer Thoughts Reviewers do more than give a rating. They also provide comments. Comments consist of a short title and a written review. Review Titles The title contents generally reflect the overall experience and, in a sense, attempt to summarize or highlight parts of the longer written review. The titles of Bloedel Reserve reviews had a median length of 23 characters (Figure 18). An example of a 23 character title is “Lovely house and garden�.

80

Number of Reviews

median title length: 23 characters

60

40

20

0 0

20

40

Title Length (characters)

60

Figure 18. The frequency of title lengths of Bloedel Reserve reviews, in characters.

37


Some titles were much longer, about the length of this line of text. A few were very short. An example of a short title is “Magical”. Title Words It is the words used in the titles that are most revealing. The title word cloud (Figure 19) shows most of the words used in the titles. A few common words (such as “is,” and “and”) are not included. Obvious words, such as “Bloedel” and “Reserve” are also omitted. Words used only once or twice are also skipped. Singular vs plural terms have been collapsed into a single term; for example, “gardens” and “garden” are all analyzed as “garden.” Overall, the title words are very positive. Scan these words, remembering that the size of the word indicates its frequency of use. You get a good sense of the character of Bloedel Reserve and the experience that a visitor is likely to enjoy.

jewel

oasis

afternoon

gem little

see

beauty

best relax

spot back

park

pleasant

time

away

grounds

walk scenery

like

natural

peaceful

beautiful outing

location

simply

woods way

one

delight

quiet

spend experience

setting

stroll

hike

explore plants

gorgeous place garden summer

even get

year

landscape

green

seattle

spectacular

must nice tranquil great day well easy stunning

places

38

perfect

lovely visit nature worth serene

northwest

kept

maintained

wonderful

beautifully calm

tour

go peace

favorite

magical

enjoy

incredible

find

amazing

love

spring

Figure 19. The words used in the titles of reviews of Bloedel Reserve over the period of 2012-2018. A few very common words are not included and some words are treated as synonyms. Only words used more then 2 times are included.


39


It is also interesting to look for “missing” words. “Color” and “water” are two examples. There are spots of color and lots of water in the Reserve. But these sorts of terms (except, perhaps, “green”) have not made it into the set of title words. Written Reviews The written reviews provide information similar to the title, but with much more detail. Figures 20 and 21 give the lengths of the written reviews (sometimes called “comments” here). One view is from the perspective of characters and the other from sentences. A review with a median length of nearly 350 characters would take an average American typist (35-40 WPM) a little less than two minutes to complete. The longer reviews, some reaching over 2,000 characters, would take about ten minutes to type. This time doesn’t include the period needed for thinking about the review and proofing the typing before it is submitted. Long reviews require substantial effort. The usual-length review for the Bloedel Reserve also takes appreciable time. Doing a review, especially a long one, is an interesting measure of commitment.

40


50

median comment length: 349.5 characters

Number of Reviews

40 30 20 10 0 0

1000

2000

Review Length (characters)

Figure 20. The frequency of review lengths of Bloedel Reserve reviews, in characters.

median sentence length: 4.5 sentences

Number of Reviews

75

50

25

0 0

10

20

Review Length (sentences)

30

Figure 21. The frequency of review lengths of Bloedel Reserve reviews, in sentences.

41


The words used in a written review provide the message, similar to the review titles. The written reviews are more expansive and generally include more facets about the visits than the titles. The written review word cloud (Figure 22) was created with a procedure similar to the title word cloud; common or unuseful words were omitted andsimilar words were collapsed. The minimum frequency was set higher at 16 words. There are many similarities between the two word clouds. That’s expected. What creeps in, however, are lots more details. Here, “house,” “ferry,” and “japanese” are common words. Like the title word cloud, the review word cloud gives an interesing overview of a visit to the Bloedel Reserve.

42


scenery

entrance

across

variety

several

family first

pond

times

stunning

part

interesting

bring

small

relax give

tour

manicured

found

absolutely

spectacular

hour

stroll flowers

minutes live

need

ride

find

moss natural us worth area now

great car hours old

nature

think

best

short

truly

explore

right

years

property along per

forest

main

seattle peaceful

enjoy plants trees view estate two even day

summer

map

hike amazing tranquil

love

couple

walk visit just staff

acres

spring sit

park friendly perfect

mansion

much

spend

got

terminal

many time one also take sure place serene

little

stop

setting

admission

lovely

garden well bit

long

special

way get

lots

every lot

different

public

went experience

trip

see beautiful house around woods

seen

fee

nice

saw

come

history make

must japanese ferry areas like trail miles path can good back grounds go wonderful shop

large away

taking

home

able

easy year really sound quite beauty bus open water puget

enough

weather

person

people

definitely miss

residence

meadow

season

afternoon

recommend

visitors northwest

fantastic

gorgeous

beautifully

though

something

maintained quiet highly meadows

shuttle

gift spent want

going

mile

takes

Figure 22. The words used in the comments of reviews of Bloedel Reserve over the period of 2012-2018. A few very common words are not included and some words are treated as synonyms. Only words used more than 16 times are included.

43


Words in Less Favorable Reviews The words were extracted from the written reviews where the rating was Average (3) or Poor (4). This was done to see if there were terms that might identify specific problems or shortcomings. The word cloud (Figure 23) required at least two uses of a term to be included. As with the other word clouds, some common words were deleted and synonyms were collapsed. It is interesting that negative words did not appear in this word cloud. Note that there were very few less favorable reviews so this analysis may not be very informative.

44


long

hours explore

beautiful

view japanese

enough top place time like

us garden take seats lot

area

walk one visit park

really house however opportunity

sound

Figure 23. The words used in the comments of reviews by people who rated Bloedel Reserve less than 4 (i.e., poor or average) over the period of 2012-2018. A few very common words are not included and some words are treated as synonyms. Only words used more then 2 times are included.

45


Sentiment Analysis Many English words have a sentiment value. That is, they can be classed as either positive or negative terms. Compilations of sentient words can be used to give an indication of a review’s positive or negative character. The first analysis here looks at the written reviews based on the sentient value of their words. A score of zero is a neutral review. Increasingly larger scores indicate a greater preponderance of positive words. Negative reviews have negative scores. A sentient score was determined for each of the written reviews. The frequencies of the individual scores is shown in Figure 24. The red vertical line in this plot shows the location of neutral sentiment. The strong use of positive words is not a surprise. This is a mirror of the average review ratings. Only a few reviews are composed of generally negative words. Similarly, there were only a few average or poor ratings.

46


80 Negative

Positive

Frequency

60

40

20

0 0.0

0.2

0.4

Review Sentiment Words

0.6

Figure 24. The frequency distribution of the sentiment word score (based on the use of positive and negative words) from all reviews about Bloedel Reserve, 2012-2018.

47


Relating Sentiment to Ratings We can dig deeper. Figure 25 shows the sentiment score for the written reviews in each of the rating categories. You might expect those people who give a low rating (2 or 3) to generally use negative words in their written review. This is not the case. Similarly, the written reviews from the higher ratings (4 or 5) don’t contain many more positive words than the lower rated reviews. What does appear to differ is the range of sentiment in the higher rated reviews (4 and 5). However, a word of caution is in order. Recall that the number of lower rated reviews is much smaller than the higher ratings. Therefore, the range of sentiment values might simply be due to the relative amount of data in each rating category.

48


Review Sentiment Words

0.4

0.2

Positive

0.0 Negative

2

3

Rating

4

5

Figure 25. Statistical distribution of the positive and negative words in the reviews about Bloedel Reserve, with the analyses grouped by a reviewer’s rating, 2012-2018.

49


Review Writers We don’t have much specific information about who is writing each review. The reviews are anonymous. There is some general information about each reviewer. Reviewer Gender We can infer the gender of each reviewer’s TripAdvisor account. This is, presumably, the person writing the review. We also know something about how many reviews they have written along with the number of “thumbs up” given by readers. Overall, females wrote nearly twice as many reviews of the Bloedel Reserve as males. Figure 26 shows the proportion, broken down by the number of reviews for each of the rating values. It is easier to see the relationship between how men and women view the Reserve by normalizing the ratings as a percentage of the ratings for each gender (Figure 27). The total of lower rated reviews (2 and 3) is about the same between men and women. Men, however, seem to give slightly more Excellent (5) ratings, on average, than women in their review of Bloedel Reserve.

50


200

150

Ratings

Rating 2

100

3 4 5

50

0 f

Gender

m

Figure 26. The number of each rating for Bloedel Reserve, by gender, for 2012-2018.

100%

75%

Ratings

Rating 2

50%

3 4 5

25%

0% f

m

Gender Figure 27. The percentage of each rating for Bloedel Reserve, by gender, for 20122018.

51


Reviewer Productivity and Helpfulness Many people write many reviews. The Bloedel Reserve reviewers have written, on average, about 50 reviews (Figure 28). This shows considerable effort to make the TripAdvisor review system useful. Some reviewers are particularly active. A few have written over 1,500 reviews. Writing a review is one thing. Getting a positive response for a review is another. This positive response means that someone has found a review that is helpful. The Bloedel Reserve reviewers have gotten, on average, about 62 “thumbs up” (i.e., helpful clicks) for their reviews (Figure 29). Note, this is for any review in their set of reviews, not just their review of the Reserve. There is no measure of whether a specific review has gotten a “thumbs up.”

52


Number of Reviewers

100

Median Reviews Written: 50

75

50

25

0 0

500

1000

1500

Reviews Written

Figure 28. Total number of reviews written by Bloedel Reserve reviewers. Data from all reviewers during 2012-2018. Median Helpful Reviews: 62

Number of Reviewers

75

50

25

0 0

50

Helpful Reviews

100

Figure 29. Number of helpful votes given to Bloedel Reserve reviewers. Data from all reviewers during 2012-2018.

53


This “helpfulness” measure would seem to translate into the reliability of a reviewer’s contributions. A plot of the reviews written vs the helpful reviews (Figure 30) shows what appear to be two groups of points. There is a tight clustering along a 1:1 line. This represents reviewers that have received the same number of “thumbs up” as the number of reviews they have written. Remember that a reviewer can get many “thumbs up” for a single review while getting none for other reviews. This line is a loose and arbitrary index of helpfulness. The second group of points falls well to the right of this line. These are reviewers who tend to write a lot of reviews but fail to get lots of “thumbs up.” It is likely that these reviewers are, in general, likely to be less reliable. It might be useful to keep this relationship in mind when reviewing detailed comments. A reviewer who scores much lower than 1 on this helpfulness index might not have much credibility among his or her peers.

54


125

Reviews Helpful

100

75

50

25

0

● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ● ●●● ● ● ● ● ● ● ● ● ● ●● ● ●● ●●● ●● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ●● ● ●●●● ● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ●● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ●● ●●● ●● ● ●● ● ● ●●●● ● ●● ● ● ● ●● ● ● ● ●●●●● ● ● ● ●●● ● ● ● ● ● ● ●● ●● ● ● ●● ● ● ● ● ● ● ● ●●● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●●●● ● ●●●● ● ● ●● ●● ● ●●●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ●

0

500

● ●

● ●

● ●

● ● ● ●

1000

Reviews Written

1500

Figure 30. Relationship between the number of reviews written and the number of helpful reviews. Red line shows the 1:1 relationship. All Bloedel Reserve reviewers over 2012-2018.

55


Reviewer Origins TripAdvisor reviewers are, most often, listed with their home location. Knowing the origin of visitors is important. For the Bloedel Reserve, in-state visitors make up nearly 42% of the visitors with a known home location (Figure 31). We’ll revisit this topic later (starting on page 66) as there are many ways to visualize, and understand, the reviewer home locations.

Washington Visitors: 41.8 % Reviews

200

100

0 Outside Washington

56

Washington

Unknown

Reviewer Origin

Figure 31. The number of Bloedel Reserve reviewers from inside Washington, compared to elsewhere, 2012-2018.


57


When Reviewers Visit The month a review was written is assumed to be the same as the month of the visit to the Bloedel Reserve. Reviewers visited the Reserve (Figure 32) with a peak in July, with August coming in as the month with the second most visitors. Note that there are a lot of visitors besides the reviewers. They may be local residents who come at different times, some likely enjoying the Reserve several times each year.

58


January

Note: Only USA

February

Month of Review

March April May June July August

September October November December 0

20

Visitor Reviews

40

Figure 32. The number of reviews each month for Bloedel Reserve.

59


The two month peak reviewer visiting period, not unexpectedly, is when the weather is drier at Bainbridge Island (Figure 33). This plot is not the monthly rainfall, but the rainfall probability. That is more realistic measure of how people decide on their outdoor activities. There is more to the decision process. Figure 34 shows a monthly desirability index based on temperature, clouds and precipitation. This is what weatherspark.com calls its “tourism index.� The two months, shown in green, are considered the best times to visit Bainbridge Island.

Wetter

Drier

Wetter

Rainfall Probability

60

40

20

0 J

F

M

A

M

J

J

Month

A

S

O

N

D

Figure 33. The general probability of rainfall at Bainbridge Island for each month. Redrawn from weatherspark.com.

60


Desirability Index

6

4

2

0 J

F

M

A

M

J

J

Month

A

S

O

N

D

Figure 34. A composite score, largely based on environmental variables, showing the relative desirability of visiting Bainbridge Island in each of the months. Redrawn from weatherspark.com.

61


Both the monthly reviewer visits (Figure 32) and the Desirability Index (Figure 34) show similar general distributions. They each have peaks in the summer and low points in the winter months. The relationships between these monthly values are shown in Figure 35. This is a linear regression where the Desirability Index is used to predict the Number of Reviews. The data points closely follow the line of best fit. Technically, the r2 value is 0.88 (P<0.01). That’s a very good relationship. Weather is, indeed, a predictor of reviewer visits.

62


50 ● ●

Number of Reviews

40 ●

30 ●

20 ● ●

10

● ●

0

2

4

Desirability Index

6

Figure 35. The monthly relationship between desirability, which combines several environmental factors, and the number of visitor reviews, for 2012-2018.

63


Persistent Reviewer Attitudes Each TripAdvisor review includes the distribution of review ratings for all of the reviews submitted by the reviewer. For example, a reviewer might have done 10 reviews, with 4 rated Excellent, 3 rated Very Good, 2 rated Average, and 1 rated Terrible. That information is available, but a bit difficult to dig out. The question is whether a reviewer tends to always give the same ratings, independent of the attraction being reviewed. For example, if someone has ranked Bloedel Reserve as Poor, has that reviewer generally given a Poor rating for that person’s other reviews? Similarly, do Excellent ratings come from people who tend to always give an Excellent rating? Figure 36 shows the distribution of all ratings for reviewers who rated Bloedel Reserve in each of the four categories. Note that there are relatively few reviewers who gave Bloedel Reserve a Poor or Average rating. That means we need to be cautious in our interpretation. With that understanding, it appears that there is no persistence of attitude relative to the rating categories. Someone who gave the Bloedel Reserve an Excellent rating seems as likely to give some other review a Poor or Terrible rating as a person who rated the Reserve Poor or Average. That’s a good sign. It indicates that you can place more trust in the reviews.

64


All Review Scores Distribution

100%

75% Rating Excellent Very Good

50%

Average Poor Terrible

25%

0% Poor

Average

Very Good

Excellent

Reviewer Rating of Bloedel Reserve

Figure 36. The distribution of all ratings made by reviewers of Bloedel Reserve over the period from 2012-2018. The rating distribution is the percent of the ratings made by people in each rating category.

65


US Reviewer Origins TripAdvisor reviewers are, most often, listed with their home location. Knowing the orgin of reviewers gives some insight into the visitor population characteristics. Are reviewers mostly from neighboring states? Are there many reviewers from distant states? You can see a bit of a pattern by looking at the state-by-state data on a map (Figure 37). Note that a grey-color state had no reviewers. As before, because one value (in this case, the number of reviewers from Washington) is so much larger than the other values, the relationships between the other states is obscure. Only a few other states stand out.

Reviews

100

50

Figure 37. The number of reviews of Bloedel Reserve from each state, 2012-2018.

66


The distinction between the states is seen more clearly when the log values of the number of reviews is used (Figure 38). Many states, including those on the East Coast, now stand out as having a reasonable number of Bloedel Reserve reviewers. There is still a bias in these maps. States with larger populations, like California, are likely to have more reviewers. This population bias can be corrected.

Reviews (log10) 2.0 1.5 1.0 0.5 0.0

Figure 38. The number of reviews (log scale) of Bloedel Reserve from each state, 2012-2018.

67


Figure 39 shows the number of reviews (using a log scale) per million populaton for each state. Now, smaller neighboring states, like Oregon and Idaho, are seen to have an appreciable number of reviews, given the size of the state’s population. Figure 40 separates out the states with high numbers of reviewers, again corrected for population size. Interestingly, Hawaii and South Dakota are leaders, behind Washington, for contributing reviews of Bloedel Reserve. The national picture hides the patterns within the states. The next task is to look at the reviewer distribution for the state of Washington.

Reviewss/Million (log10) 1.0 0.5 0.0 −0.5

68

Figure 39. The number of reviews (log scale) of Bloedel Reserve per million state residents, 2012-2018.


Idaho

Note: Log axis for Reviews per Millon Residents

Massachusetts Oregon

Reviewer State

North Dakota Connecticut Colorado California New Mexico Delaware South Dakota Hawaii Washington 1

3

10

Reviews per Million Residents

Figure 40. The number of reviews (log scale) of Bloedel Reserve per million state residents, 2012-2018. Only the top states are shown.

69


Washington Reviewer Origins Extra effort was devoted to extracting a reviewer’s city for Washington residents. These cities were geocoded and combined to get the distribution of the number of reviews across the state. The state-wide pattern is shown in Figure 41. Most of the reviews come from people living in the general Seattle-Tacoma area. We can zoom into the Sea-Tac area and see the more detailed distribution of reviewers (Figure 42).

Freq 10 20 30 40 50

70

Figure 41. The origin of reviewers living in Washington visiting Bloedel Reserve, 20122018.


Freq 10 20 30 40 50

Figure 42. The origin of reviewers living close to Seattle visiting Bloedel Reserve, 2012-2018.

71


Getting to Bloedel Reserve The straight-line distances between the Washington review locations and Bloedel Reserve were calculated. On average, a reviewer traveled almost 18 miles to get to the Reserve (Figure 43). There is a cluster of distances that shows most of the Washington reviewers travel 50 or fewer miles.

Number of Visitors

15

Mean Distance: 17.9 miles

10

5

0 0

72

50

100

Distance from Bloedel Reserve (miles) Washington State Residents

150

Figure 43. The straight-line distances to Bloedel Reserve for people who live in Washington.


73


Part III: Bloedel Reserve as a W


World-Class Botanical Garden


USA Today Best10 Botanical Gardens USA Today conducts an annual, on-line survey each year that allows participants to vote on their favorite botanical garden. This is not an open-ended survey. A panel of “experts� decides which ten gardens will appear on the annual list. On-line participants can then vote on their favorite garden on this list. Voters are limited to one vote per day. The voting period lasts one month. There are four years of Best10 botanical garden results available. Bloedel Reserve is included in the 2018 list at number 10. A total of 23 botanical gardens have been on the list over the four years. One garden, Missouri Botanical Garden, has been on the list every year. Four gardens have made it on the list for three years. Twelve (more than half) have appeared a single time. These data, along with the annual visitor numbers are shown in Figure 44. Being selected for the voting list appears to be the biggest hurdle. Visitor count doesn’t seem to be an important criterion to be included on an annual list, although the five gardens to appear on at least three lists have at least 450,000 visitors every year. Three quarters of the gardens appearing on the list only once had an annual visitor count below 450,000. Note that Bloedel Reserve has the lowest annual visitor count of all of the Best10 Botanical Gardens.

76


Figure 44. List of the gardens included in the 2016-2019 Best10 Botanical Gardens listing by USA Today. Visitor data from Wikipedia or garden websites. ID

Garden

2016 2017 2018 2019

Visitors

1

Atlanta Botanical Garden

.

9

6

7

500,000

2

Bloedel Reserve

.

.

10

.

30,992

3

Bok Tower Gardens

.

.

.

3

162,665

4

Brookgreen Gardens

.

.

9

.

360,000

5

Cheekwood Estate and Gardens

9

.

.

6

225,000

6

Chicago Botanic Garden

8

10

.

8

1,074,000

7

Coastal Maine Botanical Gardens

6

.

.

.

200,000

8

Dallas Arboretum and Botanical Garden

7

.

.

.

979,219

9

Denver Botanic Gardens

.

.

5

.

1,322,507

10

Desert Botanical Garden

10

4

7

.

450,578

11

Fairchild Tropical Botanic Garden

.

.

8

.

200,000

12 Frederik Meijer Gardens & Sculpture Park

5

8

.

.

185,966

13

Lewis Ginter Botanical Garden

4

.

.

4

350,000

14

Longwood Gardens

1

.

1

.

1,530,237

15

Mendocino Coast Botanical Gardens

.

5

.

.

70,000

16

Minnesota Landscape Arboretum

.

1

2

1

500,000

17

Missouri Botanical Garden

3

3

3

2

1,037,216

18

Montreal Botanical Garden

.

.

.

10

900,000

19

New York Botanical Garden

2

2

.

.

1,300,000

20

San Diego Botanic Garden

.

.

.

9

232,000

21

State Botanical Garden of Georgia

.

7

.

.

230,000

22

The Dawes Arboretum

.

6

.

.

250,000

23

Vallarta Botanical Gardens

.

.

4

5

40,000

77


Best10 Botanical Garden Locations The spatial distribution of the 10Best Botanical Gardens on the four years of annual lists seems to be important. Figure 45 shows the locations of the Best10 Botanical Gardens. Two of these gardens are outside the US. One is in Canada (Montreal Botanical Garden) and the other in Mexico (Vallarta Botanical Gardens). The gardens are generally spread out across North America. It is useful to compare this distribution to the locations of US botanical gardens listed in Wikipedia (Figure 1). It appears that candidate gardens are, in part, selected to represent different parts of North America. This doesn’t explain why some gardens, such as the outstanding Huntington Botanical Garden, is not included and the much smaller San Diego Botanical Garden is the only Southern California representative. In spite of what might be serious limitations, the comparison of the 23 gardens is interesting.

78


50

2 ● 16 ●

40

9 ●

15 ●

10 20 ● ●

8 ●

30

18 ●

7 ● 12 ● 6 ● 19 22 14● ● ● 17 ● 13 ● 5 ● 21 1● 4 ● ●

3 ● 11 ● 23 ●

20 −120

−100

−80

Figure 45. Locations of the 23 gardens featured in the annual USA Today Best10 botanical gardens, 2016-2019.

79


Rough Rating The annual Best10 Botanical Garden competition gives the order of the gardens for a single year of competition. There are several ways to evalute the composite of the annual lists. It was mentioned earlier that you can count the number of years that a particular garden has been on an annual list. Missouri Botanical Garden has the distinction of being the only garden to be on all four lists. There are five gardens that have been on three or more lists. Besides Missouri, they are Chicago, Minnesota, Atlanta and Phoenix’s Desert Botanical Garden. Note that four of the five are in the middle of the US (mostly midwest). It is possible to combine the years to include the annual ranking. Here is the procedure. The top ranked garden gets 10 points. Each rank below gets one less point. The bottom ranked garden gets a single point. Add up the annual points for the four years. The garden with the highest number of points is the top garden; it is ranked first overall. The results of this ranking are shown in Figure 46. As expected, Missouri is the top ranked garden. Bloedel Reserve (red bar) is tied for last place with Montreal. It is worth noting that the top five gardens (green bars), when viewed from the number of years on the list, are spread out in this more detailed ranking. These rankings are entirely from the Best10 data. What would we learn if we combine these data with TripAdvisor ratings?

80


Montreal Bloedel San Diego Brookgreen Fairchild Georgia Dallas Dawes Maine

Garden

Mendocino Denver Chicago

Cheekwood Bok Tower Meijer Atlanta Desert

Better

Vallarta Ginter New York Longwood Minnesota Missouri 0

5

10

15

USA Today 4−Year Rating

20

Figure 46. Relative ranking of each of the USA Today Top 10 Gardens using a combination of the annual rating for each of the years, 2015-2018. Green bars are the five gardens appearing in three or four annual lists.

81


Best10 Combined with TripAdvisor TripAdvisor reviewers have rated all of the gardens in the Best10 list. Instead of using the USA Today annual values, the gardens can be compared using the average of the TripAdvisor ratings. The TripAdvisor website gives rating details for each garden. This consists of the number of reviews for each of the five rating values. An overall rating index can be constructed from combining these five values, with each rating being weighted. For example, ten Excellent ratings would be worth 50 points while 10 Terrible ratings would give a score of 10 points. The total points is divided by the total number of reviews. This gives what is called the Adjusted TripAdvisor Rating. Note that TripAdvisor provides a similar overall rating but with less precision. The procedure used here is more precise as all of the gardens are highly rated and small average differences separate the contenders. These differences are likely to be significant because of the large sample size (i.e., number of reviews for each garden). Using the TripAdvisor overall ratings, Bloedel Reserve has moved much higher in the ranks of the 23 Best10 gardens. With this scale, it is well ahead of three of the top five based on the USA Today voting (Figure 47). This high ranking is a considerable achievement, especially considering that Bloedel Reserve has the smallest annual visitor count. This also reinforces the importance of TripAdvisor reviews.

82


Cheekwood Montreal San Diego Vallarta Bok Tower Atlanta New York Desert Denver

Garden

Georgia Dallas Ginter Minnesota Mendocino Dawes Meijer Fairchild

Better

Bloedel Longwood Brookgreen Chicago Missouri Maine 0

1

2

3

Adjusted TripAdvisor Rating

4

5

Figure 47. Adjusted TripAdvisor ratings, based on the rating frequencies, for each of the USA Today Top 10 Gardens, 2015-2018. Green bars are the five gardens appearing in three or four annual lists.

83


Encouraging TripAdvisor Reviewers TripAdvisor reviews have been shown to be an important source of information. Most institutions that are reviewed are likely neutral about whether or not a visitor submits a review. A few attractions are very aggressive (as any frequent traveler will know); they specifically ask visitors to submit a review. Bloedel Reserve is one of the neutral places. There seems to be no obvious effort to solicit TripAdvisor reviews. It would be useful to know the percentage of visitors that submit a TripAdvisor review. We can do this, at least approximately, for Bloedel Reserve because we tabulated the annual number of reviews. Divide that by the annual visitor count (something from Bloedel Reserve information) and you get the percentage of visitors doing reviews. This would be approximate for Bloedel Reserve as we only have an average annual attendance, not year-by-year values. Comparisons to other gardens would require going through tens of thousands of TripAdvisor reviews (as this study did for about 400 Bloedel Reserve reviews). And it would, ideally, use actual annual visitor counts. That is impractical. Instead, the total number of TripAdvisor reviews (collected over the entire period that TripAdvisor has been collecting reviews) is divided by the current annual visitor count. This is called the “Reviews Index.” This index is not an average, of course. But it is an index that should be comparable for all the gardens. Those data are plotted in Figure 48, comparing the number of annual visitors to the Reviews Index. Each garden has its number (See Figure 44) in a circle. The higher the circle, the higher the percentage of visitors who do a review. Gardens in the red zone have huge attendance. They don’t need TripAdvisor reviews to attract visitors. People come to these gardens for other reasons. This is probably true for gardens in the green zone. They, too, get a lot of visitors. Most of these visitors come for reasons other than reading 84


TripAdvisor reviews. The blue and tan zones are the small-attendance gardens. These are facilities that will benefit most from TripAdvisor reviews. Gardens in the tan zone could probably increase their visibility by encouraging reviews. Gardens in the blue zone are likely doing fine. Bloedel Reserve (red 2) has the fourth highest Reviews Index. That’s interesting as many visitors seem to be motivated to share their experience at the Reserve. Garden 23 (Vallarta Botanical Gardens) must offer encouragement for reviews. Otherwise, it is hard to explain this outlier.

23

Reviews Index

0.06

0.04

Better

0.02

2 15

312 7 11 5 21 20 22

0.00 0

10 4 13

1

18 8 17

16

500,000

6

1,000,000

Annual Visitors

9 19

14

1,500,000

Figure 48. The Reviews Index is the number of TripAdvisor reviews divided by the annual number of visitors for the USA Today Best10 Botanical Gardens, 2016-2019. A high value indicates that visitors post comments about the garden relatively frequently. The colored backgrounds delimit zones discussed in the text.

85


Part IV: T


The Data


It’s About Data The primary objective of this experiment was to create a framework for a better understanding an institution. The focus was on the Bloedel Reserve, especially as it is a representative of botanical gardens in general. The major constraint was that all data must be from open-source resources. This means that anyone can find the same data on the Internet. Even more important, similar data can be used following the general protocols shown here, to analyze other non-profit organizations. This includes other botanical gardens, but is not limited to this type of institution. The primary sources: There were two primary sources of data. TripAdvisor posts reviews of Top Attractions in places throughout the world. This is a rich font of information. Wikipedia accumulates lots of useful information, too. It is particularly good for lists of things. You can find information about specific places as well. More specialized sources: Google searches turned up lots of important. These were more specific bits of data. Individual websites and links to specialized documents (such as newsletters) helped fill in data about specific institutions. Other sites let me look up information such as geographic coordiantes of zip codes and USDA Hardiness Zones. Sometimes, you need to search the Web deeply to find what you need. Persistence helped. Tying it together: It is a big step from data to a form appropriate for presentation. Fortunately, there are good tools that help. Spreadsheets were used for much of the early data entry and conditioning. As soon as possible, the files were moved into the RStudio environment. This data analysis system has tools (think: libraries) that do the heavy part of data organization and visualization. Maps and charts were exported from RStudio. The text, photos and figures were organized with Adobe InDesign.

88


There were a lot of challenges along the way. Here are a few lessons learned. Organization is very important. Create a modular structure and stick with it. If you need to “play,” create a sandbox. Try to be consistent. There are often many ways to solve a problem. Here, the ggplot2 library was used whenever possible, even through there are simpler ways of doing some of the tasks. At first, it’s hard to identify ways to be consistent. There are competing strategies for almost everything in the R language universe. Be observent of details. Establish standards. Write code consistently, including comments and typographical standards. Centralize data storage. All the data should be kept in a separate location (i.e., folder). Use a section of the RStudio project to document each data source and any basic manipulations. Provide a script to read each data file and run a simple summary analysis to check on the integrity of the contents. Examine and rerun analyses. It is rare for the first run of an analysis to be adequate. Data adjustments, supplemental annotations, format changes, and stylistic modifications are often (always!) necessary. The important thing is that these modifications should fit into the analysis protocol. Design analyses so the code for a particular analysis can be run over and over. Expect that you will add, modify and delete data. You’ll also add annotations, modify the ancillary information and change the visual characteristics. Make it a straight-forward process with careful design and planning.

89


Data Sources & Organization The Garden Table The overall goal is to have a comprehensive list of all US botanical gardens, along with a rich set of information about the gardens. This allows Bloedel Reserve to be placed in a national perspective. The dataset, called the “Garden Table,” started with Wikipedia. A list of US botanical gardens, which forms the base of the garden table, is found at https://en.wikipedia.org/wiki/List_of_botanical_gardens_and_arboretums_in_the_United_States Separate Wikipedia pages were needed for states that have a large number of gardens. Data from all of the pages were copied and transferred to a spreadsheet. This resulting list gives the garden name, location (city, state), and institutional affiliation. There are some distinct biases in this list as it appears to be synthesized from lists produced by each state. It is unlikely that each component list is based on the same inclusion criteria. Also, some compilation efforts appear to be quite different from the norm. Recognizing these limitations, these data are are used for our “big picture” analysis. Note that the gardens in Guam, Puerto Rico and the Virgin Islands are in the Wikipedia list but not used here. Garden Size

90

This Garden Table was expanded during the data entry process to include the area of each garden (acres). These data were obtained by searching garden websites and other sources (e.g., annual reports) to get as complete a set as possible. Finding the size data wasn’t easy or fast.


Zip Code for Location Garden locations are given in Wikipedia only to the resolution of the city name. The Zip Code for each city was added to the garden table. USDA Hardiness Zones The location Zip Codes, in turn, were used to determine the USDA Hardiness Zone of each garden. The lookup involved a search of the database found in the website https://garden.org/nga/zipzone/ This value, collected one at a time, was added to the Garden Table. The map of Japanese Hardiness Zones was adapted from a link on the website https://www.plantmaps.com/ There are many other useful environmentally-oriented resources on this site. City Geolocation The Zip Codes were used with Google maps to get the location of each city having a botanical garden. This involved the creation of a new data field (address) which was the combination of the city and state. With this field, it was an automated process of getting the location coordinates (whew!). Overall, creating the Garden Table was a very tedious process. However, it is a very rich source of information about US botanical gardens.

91


Trip Advisor Reviews TripAdvisor is a web-based system that allows people to review places they have visited. These reviews are used by other visitors to decide where they should visit. Reviews can serve another purpose. A reviewed location, like Bloedel Reserve, can obtain information about visitor impressions, where the reviewers live, and when they visited. Reviews are added by visitors on a voluntary basis. Some people do this regularly. There is no benefit in doing this other than contributing to a “crowd sourced” set of impressions that might benefit the community. People adding comments have registered with TripAdvisor and contributed a few details about themself, such as their name (ususally a pseudonym), and home location. Trip advisor keeps track of the number of reviews each person has added to the system. If someone using TripAdvisor feels that a review has been helpful, they can click on an icon to record their feelings. The total of these clicks (“thumbs up”) is also recorded for each reviewer. The point is, there is quite a bit of metadata about reviewers. Each review consists of three basic parts. A rating is given using a five point scale, with 5 being the highest value. The review gets a short title. The written review itself is free-form and can be as long as needed. Lots of people find places to visit through TripAdvisor. Once a location (such as Bainbridge Island) is provided to TripAdvisor, it is possible to get a set of activities (called “Things to do”). This set is ordered by an activity’s rating value, with the highest scoring activities appearing at the top of the list. When you view the information for an activity, you see a short list of re92


views. The most recently added reviews come first. It is possible to sort this list, but here, we’re concerned with using a large sample of the entire set of reviews. The TripAdvisor data for Bloedel Reserve consists of a substantial sample of reviews from April 2012 through September 2018. In some cases, mostly for metadata, the time period was extended through the end of 2018. Scraping the Primary Data Getting the TripAdvisor review information into a form suitable for analysis was a tedious affair. Information was gathered by copy and paste into a text file. The information in the file was “cleaned” and put into order with a lot of hand work. Eventually, the data made its way into a spreadsheet. This allowed further refinement. A few words were deleted or replaced in the reviews. This was intended to add uniformity, such as expanding abbreviations. Unnecessary punctuation was removed. The result was a file with a separate entry (i.e., spreadsheet row) for each review. Each scraped entry had the reviewer’s state of residence, country of residence, number of reviews written, number of “thumbs up,” a review title and the review text. Adding Rating and Gender Scraping the TripAdvisor screens did not capture the Ratings. Adding these data was done the hard way, by looking at each entry, evaluating the information, and typing the result into a notepad file. The gender data were obtained by looking at the name and picture in each TripAdvisor review. Any ambiguous evaluation was scored as missing data. There is a bias, of course. It is likely that reviews are the result of a discussion, perhaps between a couple. The writing of the review may not be the person who held the stronger view. Getting the rating was easy as it involved only counting the number of filled 93


circles. Still, the entire process was a bit tedious. Creating an Expanded File of TripAdvisor Data The combination of the three files (scraped data, gender and rating) make a rich set of information available for each review. A file was created that is a combination of the data. This new file was used in all of the analyses of the reviews.

USA Today 10Best Botanical Gardens USA Today’s 10Best is a web-based system that allows people to vote for their favorites. This system has a number of top 10 lists. This is a travel-oriented project and there are many types of favorites. The 10Best mission statement is: 10Best is the trusted source for online travel advice. We offer unbiased, top-rated recommendations for a city’s best sights, restaurants, clubs, and shops. Here is the process relative to botanical gardens. Each year, “experts” propose 20 gardens. 10Best participants vote to determine the top ten. The USA Today data were obtained from web searches. A variety of sites listed the annual voting results with each year’s listing ordered from 1 to 10 (1 is most popular). Four years of 10Best botanical garden data (2016 to 2019) were assembled. There were different gardens on the final list for each year. Some gardens were on several of the annual lists. Twenty-three gardens, in total, appeared on the four annual lists. The data were hand compiled into a spreadsheet with separate columns for the rank of each garden in each year. Summary columns were calculated for the frequency of appearence in the annual ratings and the highest value 94


in the ratings. An index was created to combine of the four years of rating values. This index is an approximation to the overall popularity of each garden over the four year period. Adding locations and garden attendance The geo-location of the city for each garden was determined with an on-line lookup. The latitude and longitude values were added to the spreadsheet. The most recent annual visitor count for each garden were obtained by Internet searching. Often, this number was prominently displayed on the garden’s webpage. In a few cases, it took considerable effort to find and compile the data. For example, the Vallarta Botanical Gardens required looking through each of their monthly newsletters to find the previous month’s attendance. These values were then summed. TripAdvisor review information Each of the 23 10Best Botanical Gardens has been reviewed (often extensively) in TripAdvisor. Several data values were extracted from TripAdvisor: the total number of reviews and the number of ratings in each of the five categories (1 to 5). The ratings were averaged by multiplying the number of ratings for a category by the category value. This provided an adjusted rating value that is more precise than the averaged value given on TripAdvisor. This was an important refinement as these gardens all have high ratings and differences can be detected only by using the finer scale calculated here.

95


Photo Notes Cover: Tall trees and dense understory vegetation near the Waterfall. Page 2: A path through the Moss Garden.This shows how moss, once planted as a ground cover, is now creeping up the tree trunks. Page 7: Prentice Bloedel. This is a photo of the portrait hanging above the fireplace in the living room of Collinswood. Pages 10-11. The Japanese Garden, created by Fujitaro Kubota and his sons. Page 17: A forest path Page 23: Reflection Pond Page 28: Swan Pond. It no longer has swans. The flowering tree is a Dawsons Magnolia. Pages 30-31: The lawn in front of Collinswood, the long-time residence of the Bloedels. Page 39: Moss and ferns on the forest floor in Moss Garden. Page 57: Frank Buxton Bird Marsh. This dense thicket of vegetation and open water is prime habitat for birds. Page 73: A huge fern growing near the Christmas Pond. Pages 74-75: The Japanese Guest House sits behind the Japanese Sand and Stone Garden. The bright colors of the Golden Locust tree add to the beauty of this place. Pages 86-87: Frank Buxton Meadow. 96


Endnotes 1 Kreisman, L., 1988, The Bloedel Reserve: Gardens in the Forest. The Arbor Fund. The quotation in this book comes from the University of Washington Arboretum Bulletin (Spring 1980). 2

Ibid.

3

Ibid.

97


Issuu converts static files into: digital portfolios, online yearbooks, online catalogs, digital photo albums and more. Sign up and create your flipbook.