↑ Return to Course Material

Prediction Profit Project

The Project:

Your client is the marketing division within wireless phone operator (pretend DTAC). They wish to understand who their present customers are and they wish to increase market share. The way in which they wish to increase market share is by targeting potentially high revenue customers.  You will have to predict phone plantype and also predict phone usage to the best of your ability.  Then combine the utilize those predictions to predict revenue for the prospect customers.

 

They are open to additional suggestions. You will be given the client database and a prospect database. The prospect database is the list of potential customers, but are not customers yet. Give them a marketing strategy and discuss who they should market to and how you came to your results. This is to be a mail out campaign, they will mail to the prospects selected by your team. This will be a group project. Groups are to work separately.

There are 3 main purposes behind this project:

  1. Learn about discriminant analysis, a multivariate technique. Within the presentation you must explain the performance of your model, how good or bad, in a manner that a non-statistician would understand. This is a very difficult task, but this will increase your understanding greatly. This will also let me know the level of understanding you have of what you are doing.
  2. Increase your knowledge of general linear models. Within the presentation you must explain the performance of your model, how good or bad, in a manner that a non-statistician would understand. This is a very difficult task, but this will increase your understanding greatly. This will also let me know the level of understanding you have of what you are doing.
  3. Give you an example of a targeted marketing strategy

The Deliverable:
You will present the project when it is due.  The presentation should be done in PowerPoint.  Graphs, etc. must be done in Microsoft Excel.  You will pretend you are from a consulting firm presenting to a client with very little to no statistical background.
Finally, you will hand in the PowerPoint presentation and all supporting materials.
The appendix should contain the statistical procedures used and the output.

You will be graded mostly on the PowerPoint Presentation, but not solely.  The presentation should be made as a stand alone document.  It should be made in a manner that I can read it without you needing to present for me to understand.

The Client Data File Contains
Name
Gender (male=1)
Home owner or not (own=1)
Cell phone plan type there are 5 plan types. The minimum minutes charged for:
1. no minutes charge 4.0 baht/minute (No minimum payment)
2. 200 minutes charge 3.5 baht/minute (Pay at least 700 baht)
3. 400 minutes charge 3.0 baht/minute (Pay at least 1200 baht)
4. 600 minutes charge 2.5 baht/minute (Pay at least 1500 baht)
5. 800 minutes charge 2.0 baht/minute (Pay at least 1600 baht)
Customer income
Government job or not (if work in government=1)
Age category (age categories go youngest to oldest)

Location (place of residence ordered, to be considered an actual address)

Minutes (the total number of minutes used for the most recent month)
Payment history. You are given the past 12 months on whether the customer paid on time or was late and how late.
1. 0=current, not late on payment
2. 1=30 days late
3. 2=60 days late
4. 3=90 days late
5. 4= in default, and are not expected to pay, very bad

Note: The Prospect Data Does Not Have As Much Information As Your
Client Database.

Phone Client Data:

  1. Phone Client data group1
  2. Phone Client data group2
  3. Phone Client data group3
  4. Phone Client data group4
  5. Phone Client data group5
  6. Phone Client data group6
  7. Phone Client data group7
  8. Phone Client data group8
  9. Phone Client data group9
  10. Phone Client data group10
  11. Phone Client data group11
  12. Phone Client data group12
  13. Phone Client data group13
  14. Phone Client data group14
  15. Phone Client data group15

Phone Prospect Data:

  1. Phone Prospect data group1
  2. Phone Prospect data group2
  3. Phone Prospect data group3
  4. Phone Prospect data group4
  5. Phone Prospect data group5
  6. Phone Prospect data group6
  7. Phone Prospect data group7
  8. Phone Prospect data group8
  9. Phone Prospect data group9
  10. Phone Prospect data group10
  11. Phone Prospect data group11
  12. Phone Prospect data group12
  13. Phone Prospect data group13
  14. Phone Prospect data group14
  15. Phone Prospect data group15