Use which arithmetic to analyze customer churn
-
Monday, March 18, 2013 3:07 AM
Hi All,
I am using SQL Server Datamining to analyze customer churn.
What I want is: I want to get provide one score to each customer based on their properties.
Take one example:
CustomerId, AgeGroup, Education,ChurnStatus
1, 20-30, Master,Churned
2,30-40, Bachelor,Active
3, 10-20, SeniorHighSchool,Churned
Could you tell me I should use which arithmetic to get how important are the properties?
I want to get this result, after using the arithmetic, then I can get one score lists for each property.
AgeGroup 20-30, 0.5
AgeGroup 30-40, 0.3
AgeGroup 10-20, 0.9
Education Master, 1.1
Education Bachelor, 1.5
Education SeniorHighSchool, 1.3
Then I can now for CustomerId, his score is 0.5+1.1=1.6
Is it possible to achieve that?
Thanks.
All Replies
-
Monday, March 18, 2013 5:08 AMAnswerer
Naive Bayes algorithm give this information. If you have never used SQL Server data mining, I recommend you to start with data mining add-ins for Excel.
Once you install the add-in and prepare your data, you can run "Analyze Key Influencers" task that will create a nice report showing exactly what you are asking for.
Note that you will need to prepare data first and add the "ChurnStatus" column. Usually it is not available in your customers database and you will need to define it (e.g. look at the transactions history and set ChurnStatus to "True" if customer has not made a purchase in the last N days).
Another task you might want to look at is Prediction Caculator. It will create a report that you can use to make predictions yourself.
Tatyana Yakushev [PredixionSoftware.com]
- Marked As Answer by BIMaster Monday, March 18, 2013 6:24 AM
.png)
