Identifying Customers who are more likely to purchase a given product category- Which model to use?

Question

Identifying Customers who are more likely to purchase a given product category- Which model to use?

Surya Gururajan

2018年11月25日 20:01

I'm pretty new to data science. I am working on a model to identify customers who are more likely to purchase a given product category.

I did try market basket using arules package from R. I also used MatchIt, Matching algorithm and did try Propensity score matching to find similar customers who tend purchase in the given product category. But I'm not quite sure the model is good enough for this kinda problem.

I'm just wondering what kind of model will be useful to find customers who are more likely to purchase a given product group. Any help is greatly appreciated.

Thanks in advance.

Topic market-basket-analysis python predictive-modeling r

Category Data Science

HonzaB · Accepted Answer · 2017年9月8日 08:12

Actually, I'd propose to tackle it from a different angle - assuming you want to predict product categories, and those are usually few, you can see it as a regular classification problem.

Every product group is a category, which you will try to predict. So you can use simple models, which are easy to train (logistic regression, decision trees etc.), easy to build, debug, explain and test. And you can experiment with various features - which ones are the most helpful when predicting product category (this is crucial, since many people will probably ask you why the model predict this category for this person).

Once you master this, then the regular approach to this is any type of recommender system (collaborative filtering, matrix factorization etc.). Plus you can use your simple approach as a benchmark.

Dinesh Kumar · Accepted Answer · 2017年8月9日 05:15

1

Dinesh Kumar answered at 2017年8月9日 05:15

Market Basket Analysis can be used. Follow this link to understand more about the algorithms used and other key concepts http://www.listendata.com/2015/12/market-basket-analysis-with-r.html

CalZ · Accepted Answer · 2017年5月11日 00:48

That is called a "recommender system" and is related to the associative rules/market basket stuff you said you tried. There is a package for it in R called recommenderlab. Here is a nice walk-through:

http://bigdata-doctor.com/recommender-systems-101-practical-example-in-r/

Identifying Customers who are more likely to purchase a given product category- Which model to use?

About