Predictive models to be used



Hi Team,

I have a few queries on predictive models as can be outlined as below:

A client from FMCG industry wants to develop a CRM capability. It has worked together with a retailer to get transaction data. Now how can the client achieve the following objectives?

  1. Develop a predictive model to predict sales in terms of demographics, city, brands etc.,
    What model (statistical) has to be used here?
  2. Predict the products that are being bought often and suggest ways to push other fewer selling products from the store
    What model (statistical) has to be used here? Like market basket analysis or segmentation etc.
  3. Predict the SKU characteristics and parameters important to consumers due to which certain products are bought over the others
    What model (statistical) has to be used here?

It would be really appreciating if you provide solutions to these problems in my hand.

Thank you in advance.



@prabhat_dash1- All these tasks are related to suprivised learning and you can use any suprivised learning algorthim to predict the ouput.

Some of the suprivised learning algorthim-
1- logistic regreesion(for classification)
2-linear regression(for regression output)
3-Decision tree

Hope this helps!



Hi @prabhat_dash1

Good point by @hinduja1234. I’d like to add that for Q2. you can use association rule mining algorithms such as Apriori or FPgrowth. They work on co-occurrence of items in the dataset. For example, if a person buys a bat, he’s very likely to buy a ball with it.