Use the attached PowerPoint (Myname_AnswersProformaPA.pptx)  for answers. Every question has a document related to the question.

AprioriAnalysisOutput.xlsx

Q1: Was a passenger’s chance of survival on the Titanic random?

State your conclusion, based on the assocation analysis. Support your conclusion with at least one screenshot. Add a brief supporting explanation. Place your explanation in the applicable slide’s notes pane. To demonstrate understanding, you may wish to talk about what you see in the screenshot.

AprioriAnalysisOutput2.xlsx
Q2: Which cluster has the most number of stores?

State your answer. Support your answer with a screenshot.

Q3: List the name of one store in each cluster.

State your answer. Support your answer with a screenshot.

AprioriAnalysisOutput3.xlsx
Q4: Explain the results displayed in the Confusion Matrix.

Support your explanation with at least one screenshot. Add a brief supporting explanation. Place your explanation in the applicable slide’s notes pane. To demonstrate understanding, you may wish to talk about what you see in the screenshot.

Q5: Pick any one asset and explain how you would travel down the decision tree to make the maintenance decision.

Identify the asset you are focusing on. Support your explanation with at least one screenshot. Add a brief supporting explanation. Place your explanation in the applicable slide’s notes pane. To demonstrate understanding, you may wish to talk about what you see in the screenshot.

Temp Anomaly.xlsx
Q6: Based on this visualization, do you think the triple exponential smoothing model is a good predictor of global temperatures over time?

State your opinion. Support your opinion with at least one screenshot. Add a brief supporting explanation. Place your explanation in the applicable slide’s notes pane. To demonstrate understanding, you may wish to talk about what you see in the screenshot.

Before answering the following questions about clustering, closely review the above reference materials. To gain a deeper understanding of clustering, you may also wish to look at additional resources on the internet. SAP has also published several videos on this topic, including:

• SAP Predictive Analysis Kmeans Cluster Analysis
• Analyzing Data using SAP Predictive Analysis on HANA

Include screenshots of supporting evidence, where you feel is it relevant. Clearly label each screenshot, and refer specifically to them within your answers (do not just do a screenshot dump, without attempting to demonstrate your understanding of the visualizations). Place any explanations in the applicable slide’s notes pane.

Q7.1: What actually is a cluster? (i.e., what does this term actually mean?)

Q7.2: What is the population of objects you are attempting to partition into clusters?

Q7.3: Which attributes did you include in your cluster analysis? Why did you select these particular attributes?

Q7.4: What is the customer behavior you are actually attempting to better understand?

Q7.5: How many ‘robust’ clusters did your analysis identify?

Please support your claim with a short supporting explanation referring specifically to evidence generated from your analysis (e.g., appropriate cluster representation).

Q7.6: Once you have formed clusters, you will want to know how they differ. Members of each cluster (cluster composition) should be similar in some respects, but different in other respects. Explain this statement.

Please support your explanation by referring specifically to evidence from your analysis (e.g., appropriate cluster representation).

Q7.7: Assuming you want other work colleagues to better understand your analysis, suggest a meaningful label for each cluster you claim to have found.

Annotate an image showing all your clusters with your chosen cluster labels.

Q7.8: An important input into the clustering analysis process is the seed value specifying the number of clusters you wish the algorithm to identify (Recall: one has the option of running the analysis multiple times with different seed values). For your finalized cluster model (i.e., the cluster model you felt explained the most variance within the data), how did you determine what should be the most appropriate number of clusters?

Q7.9: The Euclidean ‘distance’ (proximity) between objects and object clusters is clearly at the heart of understanding the proposed cluster model. In your analysis, what does the ‘distance’ between clusters actually represents?

Q7.10: How ‘good’ is your analysis, and to what degree of confidence do you have in the robustness of the clusters your analysis has identified?

Just claiming you are confident – or not – in your analysis misses the point of this question. I am not asking about your ‘feelings.’

Q7.11: Is it possible to determine how stable the clusters have been / will be over time? Comment.

Q7.12: Is there any relationship between geographical clustering of customers and the clusters you identified? Explain.

Q7.13: In what other meaningful way (suggest one) could a marketing analyst usefully segment a market other than by performing a clustering analysis?