#box plot

I love board games but probably not as much as the avid users over at BoardGameGeek. A few weeks back, I stumbled on a fun dataset over at Kaggle that compiles ratings for about 5000 games.

I wanted to see how similar or different the score distributions were for the Top 10 most prolific game designers (at least according to the BoardGameGeek data). Enter the box plot, a very informative visualization that I actually haven’t used before. 

If you haven’t seen one of these, it’s pretty simple. Each designer's scores are shown from low-to-high on the y-axis. The box then shows the 25-75 percentile of the distribution, with the red line representing the 50th percentile (or median). You also have whiskers to show further edges of the distribution and red dots for outliers beyond those edges.

So in this case, you can see a much narrower distribution for Richard H. Berg, a steady but less impressive designer (per BoardGameGeek users of course) than Uwe Rosenburg, whose game scores stretch over a much wider range.

Objectives: A novel coronavirus illness (COVID-19), a highly infectious disease, was first characterised in December 2019 in Wuhan, China. More than two million individuals have been infected with the disease, which has spread to 210 nations and territories throughout the world (confirmed). The sickness was initially discovered in India on January 30, 2020, in Kerala, in a student who had returned from Wuhan. The disease has been spreading throughout India's states. The study's main goal was to identify and classify affected districts into real clusters based on similarities within a cluster and differences between clusters, so that government policies, decisions, and medical facilities (ventilators, testing kits, masks, treatment, etc.) could be improved in order to reduce the number of infected and deceased people.

Materials and Methods: In this paper, we focused on the COVID-19-affected states and union territories of India. We used cluster analysis, a data mining technique, to complete the work. Box plots were used to examine variances among various clusters for each of the variables. For each of the variables, we used the PAST software to create a scatter plot.

For each of the variables, the results of the clustering analysis and box plot approaches were obtained. Cluster I linked to the states AP, AR, AS, BR, CG, GA, GJ, HR, HP, JH, KA, KL, MP, MH, MN, ML, MZ, NL, OR, PB, RJ, SK, TN, TG, TR, UP, UK, WB, AN, CH, DNDD, DL, JK, LA, LD, PY for verified cases. Cluster II corresponded to all Indian states and union territories for cured patients, while cluster III belonged to all Indian states and union territories for death cases.

Conclusions: The study found that the cluster I states of MH, AP, AR, DL, and KL have a high number of confirmed cases. The box plots and histogram demonstrate differences between the three cases' clusters. In various states and UTs, the trend in box plots and histograms revealed a high percentage of healed patients. It was discovered that the states in Cluster III (MH, UP, KR, TN, DL, and WB) had severe conditions that necessitated the optimization of monitoring techniques that could assist the government in improving government policies, actions, and so on in order to lower the number of infected people.

Boxplot is considered an exploratory chart that lays out data into quartiles with minimum and maximum values shown at the ends of “whiskers”.  The whiskers are an indication of the variability of the data.  Longer the whiskers the more variable the data.  In the chart above, and with all categories on the x-axis, there is a 5 number summary including, minimum, first quartile, median (second quartile), third quartile, and the maximum.  

Like the bar chart this can be represented vertically or horizontally, this choice is mostly for readability and might vary based on the density of the categories or the amount of space available for the visualization.

From a layout perspective this chart is a lot like a bar with a numeric value on the Values (Y-Axis) and some dimensional data on the Category (X-Axis).  The additional data, in the above case, is Weekday and this shows the distribution of Sales across days of the week by Product Category.  

At a glance this chart type is great for understating the data variations and the median for a given category, but is more useful when each category or data point is examined. The tooltip is a key tool for finding these details and in the above example, additional related data is added to the tooltips. 

Understand The Data using Simple Measures

In our daily lives, we collect or deal with different kinds of the data. Almost, the data exists in any business nowadays and it presents the assets of our business, for example, in the health industry, you may need to know the data about the number appointment, the number of the doctors per speciality, the staff’ performance, the bed occupations, the mortality rate, the busiest day, the…

The next step after analyzing data is to present results to the target audience. In data presentation, it is not enough for you to tell your audience about your findings; you need to go further and show them the results. This is where data visualization comes in. Presenting findings in visual format make it easy for the audience to digest the information. There are various tools for data…

今回も箱ひげ図についての記事です。Tableau ならでは！の箱ひげ図の作り方について、以下にご案内致します。 

上の図は食品カテゴリー毎のカロリーのバラツキを表現している箱ひげ図です。一般的な箱ひげ図と比較して、以下3点の特徴と良い点があります。

ヒゲが最大値・最小値のところまで伸びていない→外れ値を考慮して表現できる 

一つ一つのデータに色がついている→二つのメジャーを評価・表現できる 

データが3列に並んでいる→密集しすぎてその度合いが分かりづらい点を改善する、一つ一つのデータの視認性を向上させる 

それでは、各項目についての詳細を言及していきます。 

1.ヒゲが最大値・最小値のところまで伸びていない

これは、実はTableau のデフォルト設定です。どのように外れ値を算出しているかと申しますと、上下のひげともに箱の1.5倍以下の長さとして、もしそれを越えるようなデータがある場合それは外れ値とみなす（最大・最小値とはみなさない，ひげはそこまで伸ばさない）という挙動になっております。参考）https://mathtrain.jp/hakohige 

もし、最大値・最小値のところまでヒゲを伸ばしたい場合はもちろん設定できます。 

2.一つ一つのデータに色がついている

本件については特に取り立てる内容はなく、データセットの別のメジャー項目を色にドラッグするだけです。

3.データが3列に並んでいる

こちらは、インデックス関数を使っております。関数の概要と、式の中身は以下をご参照ください。

こちらの計算結果を列に入れて、’次を使用して計算’で特定のディメンションから、バラツキを見ている粒度のディメンション（今回の場合は食品名）を選択致します。