logistics analysis in tableau using Marketing concepts

logistics analysis in tableau

1. Logistics Regression

2. Problem with RFM Selection process is conditional on 3 predictor variable – Demographics and Psychographic variables do not have any role It is not for forecasting the behavior of new customers on whom company has no data

3. Context #1 The IRS is interested in identifying variables and factors that significantly differentiate between audited tax returns that resulted in underpayment of taxes and those that did not. The IRS also wants to know if it is possible to use the identified factors to form a composite index that will represent the differences between the two groups of tax returns. Finally, can the computed index be used to predict which tax returns should be audited?

4. Context #2 The criminologist is interested in determining differences between on-parole prisoners who have and who have not violated their parole, then using this information for making future parole decisions.

5. Context #3 The marketing manager of a book store is interested in salient attributes that successfully differentiate between purchasers and non-purchasers of books, and employing this information to predict purchase intention of potential customers.

6. Linear vs logistic Linear Regression Logistic Regression 1

7. Linear Regression X1 X2 X3 Y B1 B2 B3 Y = B1*X1+B2*X2+B3*X3

8. Logistic Regression X1 X2 X3 Y = 0 or 1 B1 B2 B3 Sigmoid Transformer You still compute B1*X1+B2*X2+B3*X3    

9. Logistic Analysis Regression analysis for discrete dependent variables We find out what factors affect the probability a discrete outcome – buy vs. no buy SPSS - Analyze/Regression/Binary Logistic While in the above we are analyzing a binary business problem, it is easy extend the analysis to a “multinomial” categorical variable

10. Steps involved to profile consumers Step 1 Run a logistic Regression with all the variables that you think are important. In logistic a variable is important if the level of significance column (p-value) is less than 0.05.

11. Interpreting the output

12. Step 2 -The results for logistic should also give you the predicted probability of purchase – Save them

13. STEP 3 Rank the customer in deciles on the basis of Probability of Purchase or Response Transform/Rank Cases Rank Cases on basis of Probability of Purchase

14. STEP 4 Find out the response rate corresponding to each decile. Analyze/Report/Case Summaries

15. STEP 5 Copy and Paste the output in excel and develop a gains chart exactly the same way did we in case of RFM analysis