Page 1 of 10 - ZOOM + Part 1 Have you done A/B/N tests in your job? Is/Was there a situation that you think your job may benefit from an A/B/B test? You can also think about some other industries or products that you are familiar with. What was the situation? How did/would you design an A/B/N test? If this was a test that were finished, what was the outcome?Part 2 In the following pages, you are going to see several visualizations. I will provide some context in every example, and some explanations (and hints) if necessary. Pay attention to the axis labels and legends. Please answer the questions as detail as you could. The purpose of this homework is to test your ability to understand visualizations, to make sense of them, and to tell self-contained stories about the visualizations/data. lam sure you are capable of doing the majority of this assignment now, before we formally cover supervised learning and visualizations next few modules. For each of the 8 visualizations, please try to answer the following questions (unless otherwise noted): (1) What type of chart/plot is it? In what situation(s) would such chart/plot be the most appropriate/useful? (2) What can you see/learn from the chart/plot? Please make sure you state an observation even when it is obvious. (3) What business decision(s) may be sensible based on the chart/plot? Why? If you do not believe the chart/plot implies any actions, explain the reason(s). (4) How would you have improved the visualization (to make it more clear or more informative)? Feel free to go beyond these questions as necessary. You can add your answers to this Word document, save it, and upload to Canvas as your homework submission. \fPage 4 of 10 ZOOM + Visualization #2: The distribution of market shares (horizontal axis, in %). The vertical axis indicates the number of markets for a given market share range. The first panel shows markets that were successful (whatever that means); the second panel shows markets that have failed. And the last panel shows the distribution of all data. Ignore the third panel (Test Data). Hint: The technical name for this chart/plot is NOT bar chart. Market Share success = 1 4000 2000 200 success = 0 100 Test Data 1000 - 500 6000 All Data 4000 2000 20 40 60 80 100 Your answer: (1) ( 2 ) ( 3 ) (4 )Page 5 > of 10 ZOOM + Visualization #3: The horizontal axis shows the ranking (1 to 19) of the cost-per-click (CPC) for nineteen search keywords in an online ad campaign, whereas the vertical axis shows the ranking of click-through-rate (CTR) for the keywords. The size of the bubble indicates the number of clicks (the bigger, the more clicks). Hint: This is a scatter plot (to answer the first question), but it is not a typical scatter plot. Figure 11: CTR vs. CPC Rankings United States United Kingdom banking careers bankruptcy bankruptcy atm rewards card credit cards atm overdraft fee debit debit interest rate wire transfer mortgage banking careers consolidate loans interest rate certificate of deposit joint account mortgage CTR Ranking CTR Ranking minimum balance overdraft fee rewards card financial services minimum balance joint account consolidate loans wire transfer certificate of deposit credit cards lawyer financial services checking account checking account insurance insurance loans loans CPC Ranking CPC Ranking Your answer: (1) ( 2 ) ( 3 ) ( 4 )Visualization #4: Automobile horsepower (horizontal axis) by different number of cylinders (4, 6, or 8). You can learn more about the dataset here: https://archive.ics.uci.edu/ml/datasets/auto+mpg Hint: The yellow line indicates the median, whereas the grey lines projected to the horizontal axis indicates the mean. 4: 78.28 + 14.5 68 78 88 6: 101.51 + 14.2 92.50 100 110 8: 158.30 # 28.3 140 150 175 60 80 100 120 140 160 180 200 220 240 ANOVA: 589.590 (p=0.000, N=391) Your answer: (1) (2) (3) (4)Page 7 of 10 ZOOM + Visualization 5: From the same automobile dataset as the last visualization. 5000 4000 weight 3000 2000 60 80 100 120 140 160 180 200 220 horsepower Your answer: (1) ( 2 ) ( 3 ) ( 4 )Page 8 of 10 - ZOOM + Visualization 6: From the same automobile dataset as the last two visualizations. 1, 2, and 3 denote different countries of origin. 5000 A3 4000 weight OO XO 3000 2000 60 80 100 120 140 160 180 200 220 horsepower Your answer: (1) (2) (3) (4)ZOOM + Page 9 of 10 Visualization #7: Kickstart projects successfully funded ("yes") or not ("no") by project type. You can learn more about the dataset here: https://orangedatamining.com/blog/2018/08/27/explaining-kickstarter-success/ no 240 .yes 220 200 180 160 140 Frequency 120 100 80 50 40 20 Design Video Apps Art Crafts Gadgets Software Type Your answer: (1) ( 2 ) ( 3 ) (4)Visualization #8 (BONUS): A (partial) decision tree to predict whether a Kickstart project was funded. Here is how you can read the tree, using the left-hand side rectangle at the second level as an example: a. \"Design or Video\": The two types of projects included in this rectangle. We know "Design" and \"Video\" denote the types of projects because the top level rectangle tells us that the data is being split by "Type"; b. "yes 78.5%, 346/441": There are a total of 441 observations in these two categories. Out of these 441 projects, 346 of them were funded (\"yes\"), which is approximately 78.5% (out of 441); c. \"Images\": The tree will split next based on \"Image," which is the number of images in included in the Kickstart project page; d. The two branches below at level 3 tells us that the split is between projects with no more than 5 images (5). no 53 9%, 535/1155 we neeqn or Video Apps. An, Crafts, Gadgets or Software vs no uses, 346/4410 51.7%, 590/7220 1M sacked Proms s 5 > 5 s 1 > 1 yes no ves 53.7%, 113/15430 87.7%, 542/5150 51 5%, 55/104 0 Pro] 0a Len Goal Type s 1554 > 15214 s 1124 so > 1124.00: 1439.10 > mom Sohware An,Cra/1suroadgexs no vs ,u no no no 1/5 52 3%, 35/51 0 75.8%, 100/1270 mass, 79/99 0 55.0%, 30/123 0 93.3%, \"\"4956 um, 21/27 0 54.9%, 50/77 0 Goal Goal my] Ba; Len Pledge Leve45 Type Pledge Levels Baked Projects Your answer: (1) This is a visualization from the decision tree algorithm. (2) (3) (4)