AvinashRdy Tracker

Re: What are the implications of using MFR binning instead of RFM binning?

AvinashRdy — Mon, 30 Mar 2015 16:39:44 GMT

Thank you for the response. This is very helpful for my research

What are the implications of using MFR binning instead of RFM binning?

AvinashRdy — Wed, 25 Mar 2015 22:05:01 GMT

Hi,

I'm dealing with data from a hotel where customers visit occasionally. Here, the recency wouldn't be much higher compared to any retail store. Hence, we decided to go with binning the data starting with Monetary value first, then binning Frequency finally binning Recency. The final index would be MFR index rather than RFM.

Please suggest / comment on using MFR instead of RFM in this scenario. mi

Thank you in advance for your help.

Regards,

Avinash

Re: Please provide suggestion for RFM using SAS EG for extremely skewed distribution.

AvinashRdy — Tue, 24 Mar 2015 16:48:37 GMT

Thank you Ray for your clear explanation. I believe this is a very good starting point. I'll start experimenting with the number of bins.

-Avinash

Re: Please provide suggestion for RFM using SAS EG for extremely skewed distribution.

AvinashRdy — Tue, 24 Mar 2015 16:16:09 GMT

Hi Xia Keshan / Ray Wright,

Thank you for your response. Unfortunately, I've SAS EG 5.1 which doesn't have the option to perform RFM scores directly. Can you suggest any other way to do this?

Please provide suggestion for RFM using SAS EG for extremely skewed distribution.

AvinashRdy — Tue, 24 Mar 2015 03:29:27 GMT

Hi All,

I've a customer data set with approximately 5 million records and data is collected based on customers from past 10 to 15 years. My target is divide the customers into RFM bins. All the recency, frequency and monitory value variables are extremely skewed.

For example, if I consider recency there are about 60-70% of the customers with recency 1 & 5-10% with recency between 2 to 5 & 2% between 5 to 10 so on..also about 0.1% above 100.

Similar case with monetory value. The monetary varies from 0 to 10,000,000 there are

about 30% of the customers who spent < $5 &

about 20% of the customers who spent between 5 to 10 &

30% b/w 10 to 100 & 20% between 100 to 1000 &

with 10% b/w 100 and 10,000 and

8% b/w 10,000 to 100,000 and so on...about 0.01% > 1,000,000

Similar scenario with Recency variable.

I need to decide how the split should be done. I've access to SAS EG. Any idea or solution is much appreciated.

Thank you so much in advance for your time.

- Avinash