Hi,
RFM Segmentation is the subject of Rules Base Segmentation by taking in to account -
- Recency (When was the Households Last Shopped)
- Frequency (How Often Households shopped)
- Monetary (Value of Households for Business in terms of Gross Profit Margin)
If it is like above then you can process with RFM Segmentation by using your transaction dataset as follow:
- Create the Visit Dimension: Calculate Av. Visit Per Week for each households
- Create the Spend Dimension: Calculate Av. Spend Per Week for each households
Now based on the above two KPIs, create RFM score by combining both the KPIs like -
RFM_Score = av.spend || av.visit;
Once you have RFM_Score variable ready, do Proc Rank and you will have rank based on Spend and Visit Dimension...
Finally based on assigned rank (by using Proc Rank), create some rules as follow:
- Maximum Rank (Say 5) are High Value Households
- Medium Rank (Say 3 to 5) are Medium Value Households
- Low Rank (Say 1 to 3) are Low Value Households.
The above procedure of creating RFM Segmentation does not include Margin Dimension as you can eassily group the households based on Frequency and Recency...You can simply calculate Margin KPI and pool more insight from that...
Hope it will be useful to your project...If you can post some sample transaction data, we can discuss more on RFM Segmentation...the above discussion is the starting in creating RFM Segmeatation...
You need to fix period through which you can refresh your RFM Model like every 4 week, 8 week, 13 weeks, etc...
-Urvish
