Hi All,
I would like to model response over time (in this case Order_Flag=1). For example, I have some customers (below)), who viewed an offer on line different dates, and they might click and order or not. What's the approach here? I could use Logistic Regression, but I guess it's not recommended for repeated measures like the data below.
Also if I want to build a logistic regression and take the max(view_page_date), I might loose some responses, for example for the customer no 001000023769, his max date is on 23Oct2013 but he has responded on 28Jul2013. Coud you please recommend the best approach to model responses in this case?
Many Thanks
customer | view_page_date | message | Click_Flag | Order_Flag | Var1 | Var2 | Var3 | Var4 |
001000023769 | 04Feb2013 | A | 0 | 0 | ||||
001000023769 | 21Feb2013 | A | 0 | 0 | ||||
001000023769 | 07Apr2013 | B | 0 | 0 | ||||
001000023769 | 28Jul2013 | A | 1 | 1 | ||||
001000023769 | 23Oct2013 | A | 0 | 0 | ||||
001000048157 | 03Jun2013 | A | 0 | 0 | ||||
001000048157 | 04Jun2013 | A | 1 | 0 | ||||
001000048157 | 12Jun2013 | A | 1 | 1 | ||||
001000048157 | 07Jul2013 | A | 0 | 0 | ||||
001000048157 | 08Aug2013 | A | 0 | 0 | ||||
001000048157 | 13Aug2013 | C | 0 | 0 | ||||
001000048157 | 06Oct2013 | A | 0 | 0 | ||||
001000048157 | 12Oct2013 | C | 0 | 0 | ||||
001000056209 | 02Jan2013 | D | 0 | 0 | ||||
001000056209 | 11Feb2013 | A | 1 | 0 | ||||
001000056209 | 12Feb2013 | D | 1 | 1 |
Hello and thanks for your question. I consulted an expert on your recurrent event situation. This is an extension of survival analysis. In this case it appears that the variable MESSAGE is a time-dependent covariate. Your situation would require some data preparation.
If the response is simply YES or NO for an order being placed, not a count of orders, then I think you could do this using the Survival node in Enterprise Miner. However, it would require that you input a fully expanded data set. The documentation for Enterprise Miner describes how to prepare your data in this manner. Repeated measures analysis would not be required in this case due to the assumption of independent censoring.
A resource to learn more about survival data mining in general, in addition to the Enterprise Miner documentation, is a video on YouTube: Introduction to Survival Data Mining - YouTube
Many Thanks Laura, that's really helpful..
Don't miss out on SAS Innovate - Register now for the FREE Livestream!
Can't make it to Vegas? No problem! Watch our general sessions LIVE or on-demand starting April 17th. Hear from SAS execs, best-selling author Adam Grant, Hot Ones host Sean Evans, top tech journalist Kara Swisher, AI expert Cassie Kozyrkov, and the mind-blowing dance crew iLuminate! Plus, get access to over 20 breakout sessions.
Use this tutorial as a handy guide to weigh the pros and cons of these commonly used machine learning algorithms.
Find more tutorials on the SAS Users YouTube channel.