Building models with SAS Enterprise Miner, SAS Factory Miner, SAS Visual Data Mining and Machine Learning or just with programming

What Modelling technique to use in order to attribute the right offer at the right time (Retail)?

Reply
Frequent Contributor
Posts: 95

What Modelling technique to use in order to attribute the right offer at the right time (Retail)?

Hi All, I had a question regarding the right modelling technique to use in SAS. I am working on a project where we would like to model the next best offer for each customer at the right time. I work for a retail company and at the moment they send multiple offers to their customers but there is no statistical or mathematical logic behind...The idea will be to develop kind of machine learning (in SAS) that will know what's the best offer ( 15% or 20% off etc ) and when is the good time to send this offer. Your help would be much appreciated. Many Thanks

SAS Employee
Posts: 11

Re: What Modelling technique to use in order to attribute the right offer at the right time (Retail)

You can use EM's link analysis node, which generates a next-best-offer table. 

You can find details in this paper:

https://support.sas.com/rnd/app/data-mining/enterprise-miner/papers/2014/linkAnalysis2014.pdf

SAS Employee
Posts: 24

Re: What Modelling technique to use in order to attribute the right offer at the right time (Retail)

Link Analysis node is a good option.

 

Other options include:

- Using the Association node and/or Market Basket node to generate frequent item sets and next best offers. (Similar to Link Analysis approach.)

- If you have Text Miner, you can use PROC SPSVD or PROC HPTMINE to generate SVD features directly from transactional/COO data, and find clusters of similar users or items using the Cluster node. You can also use procedures like DISCRIM and DISTANCE to perform other common collaborative filtering operations using these SVD features.

- Using the Random Forest node, Neural Net node or other multinomial classifiers to predict the next item a user will purchase based on sequences of past purchases or the attributes of past purchases.  

Frequent Contributor
Posts: 95

Re: What Modelling technique to use in order to attribute the right offer at the right time (Retail)

Thank You for all your answers... My problem here is little different than Market Basket Analysis. We basically have 10 offers only (Vouchers) and we want personalize the offers so each customer get the right offer - Customer A might prefer to get 20% off and customer B might prefer to get £50 off £100 for example. If my memory serves me right, MBA is useful when we have lots of items and when we want to do some cross selling etc In my case I don't know what is their Product 1, I want to find out first what is their preference 1, then preference 2 , etc.... So would it be OK to use random forest and have the target with the 10 different offers ? Target=(Offer 1, Offer 2...Offer 10)? Or is there any other smarter way? Many Thanks
SAS Employee
Posts: 24

Re: What Modelling technique to use in order to attribute the right offer at the right time (Retail)

I think what you propose with random forest is a good start, but it assumes you have labeled data for past promotions or customer behavior.

 

If you do, then you can use the predicited probabilities for each target level to rank the offers for each customer exactly as you propose.

Regular Learner
Posts: 1

Re: What Modelling technique to use in order to attribute the right offer at the right time (Retail)

Let me add a suggestion, not an answer. You can predict the NBO but you still need to act on it. With a realtime decision engine like SAS Real Time decision Manager (RTDM) you can execute at the right moment which is when your customer wants to intereact with your organization.

Predictive modeling will allow you to be able to be ready with the best offering, and RTDM can present it or adjust it within the context of the customer and the interaction if necessary.

Ask a Question
Discussion stats
  • 5 replies
  • 548 views
  • 5 likes
  • 4 in conversation