Model temporal/time sequential attributes for classification

User 5374 | 7/12/2016, 10:45:31 PM

I am trying to model Sales opportunity scoring model and facing issue with attributes that are temporal. For simplicity let us assume there are 4 stages for opportunity to close. As opportunity moves through Stage1 to Stage4 chances for wining opportunity increases.

Opportunity# Stage1 Stage2 Stage3 Stage4 Close (response column) 1 Yes No No No Lost 2 Yes Yes No No Lost 3 Yes Yes Yes No Lost 4 Yes Yes Yes Yes Won 5 Yes Yes Yes Yes Lost 6 Yes Yes Yes Yes Won 7 Yes Yes Yes Yes Won

If we use the decision tree algorithms it only recognizes Stage4 column as a feature. There seems to be no way to model probability of wining based on each stage. So if opportunity is stage 2 probability of wining will be 20-30%, if opportunity is at Stage4 then probably of wining is 95%.I want to create Machine learning model so that when opportunity moves through different stages the probability of wining increases (there many more attributes to be used on opportunity but I am using stage to highlight attributes that are sequential and change with time).

Any way to model such scenarios.

Comments

User 5391 | 7/18/2016, 8:36:32 PM

I think this sounds like a lead scoring problem, but you seem to want something more sophisticated than what our lead scoring does.