two-stage models to improve response prediction - AnalyticBridge2019-10-16T09:24:03Zhttps://www.analyticbridge.datasciencecentral.com/forum/topics/twostage-models-to-improve?commentId=2004291%3AComment%3A48688&feed=yes&xn_auth=noHi,
I agree with the Heckman…tag:www.analyticbridge.datasciencecentral.com,2009-08-14:2004291:Comment:501782009-08-14T10:32:48.117ZManishhttps://www.analyticbridge.datasciencecentral.com/profile/Manish
Hi,<br />
<br />
I agree with the Heckman model strategy as it is precisely meant to take care of the fact that "given you have take step 1 - I mean responded" we model the spend or any other variable.<br />
<br />
Thanks<br />
M
Hi,<br />
<br />
I agree with the Heckman model strategy as it is precisely meant to take care of the fact that "given you have take step 1 - I mean responded" we model the spend or any other variable.<br />
<br />
Thanks<br />
M Hi Yi-Chun,
Yes, it's pretty…tag:www.analyticbridge.datasciencecentral.com,2009-07-25:2004291:Comment:490002009-07-25T16:22:09.488ZMark Richardshttps://www.analyticbridge.datasciencecentral.com/profile/MarkRichards
Hi Yi-Chun,<br />
<br />
Yes, it's pretty much that simple, and makes that much sense. I'd say Greene's "Econometrics" is a good place to start, and if you really want a deep dive then Maddala's book. You might just search on "Heckit" or "Heckman" and "Model" to find what you need. A general search might be on "instrumental Variable".<br />
<br />
Regards,<br />
<br />
Mark
Hi Yi-Chun,<br />
<br />
Yes, it's pretty much that simple, and makes that much sense. I'd say Greene's "Econometrics" is a good place to start, and if you really want a deep dive then Maddala's book. You might just search on "Heckit" or "Heckman" and "Model" to find what you need. A general search might be on "instrumental Variable".<br />
<br />
Regards,<br />
<br />
Mark One way would be tag those cu…tag:www.analyticbridge.datasciencecentral.com,2009-07-23:2004291:Comment:486982009-07-23T07:33:32.419ZAnunay Guptahttps://www.analyticbridge.datasciencecentral.com/profile/AnunayGupta
One way would be tag those customers responding AND have high sales as 1's in your logistic model. Alternatively, build two logistic models - one to identify responders and one to identify high spending responders; do a cross-tab and set the strategy.<br />
<br />
With Best Regards,<br />
<br />
--<br />
Anunay Gupta<br />
Co-founder & Head of Analytics, Marketelligent<br />
#1251, 32nd G Cross, First Floor<br />
Jayanagar 4th 'T' Block<br />
Bangalore - 560 041<br />
<br />
+91 99452 81888 (India)<br />
+1 201 301 2411 (USA)…
One way would be tag those customers responding AND have high sales as 1's in your logistic model. Alternatively, build two logistic models - one to identify responders and one to identify high spending responders; do a cross-tab and set the strategy.<br />
<br />
With Best Regards,<br />
<br />
--<br />
Anunay Gupta<br />
Co-founder & Head of Analytics, Marketelligent<br />
#1251, 32nd G Cross, First Floor<br />
Jayanagar 4th 'T' Block<br />
Bangalore - 560 041<br />
<br />
+91 99452 81888 (India)<br />
+1 201 301 2411 (USA)<br />
<a href="http://www.marketelligent.com">www.marketelligent.com</a> Could you give me more detail…tag:www.analyticbridge.datasciencecentral.com,2009-07-23:2004291:Comment:486892009-07-23T03:38:21.920ZYi-Chun Tsaihttps://www.analyticbridge.datasciencecentral.com/profile/YiChunTsai
Could you give me more details? Thanks.
Could you give me more details? Thanks. Hi, Mark:
Do you mean that if…tag:www.analyticbridge.datasciencecentral.com,2009-07-23:2004291:Comment:486882009-07-23T03:37:36.400ZYi-Chun Tsaihttps://www.analyticbridge.datasciencecentral.com/profile/YiChunTsai
Hi, Mark:<br />
Do you mean that if the independence assumption does not hold, then I could include the estimated probability from the logistic regression in my linear regression model? If that is what you meant, then it makes sense to me to do just that. By the way, which book do you suggest me to read in detail?
Hi, Mark:<br />
Do you mean that if the independence assumption does not hold, then I could include the estimated probability from the logistic regression in my linear regression model? If that is what you meant, then it makes sense to me to do just that. By the way, which book do you suggest me to read in detail? Yi-Chun's response is a fairl…tag:www.analyticbridge.datasciencecentral.com,2009-07-23:2004291:Comment:486862009-07-23T03:21:13.748ZMark Richardshttps://www.analyticbridge.datasciencecentral.com/profile/MarkRichards
Yi-Chun's response is a fairly common approach (and a good description).<br />
<br />
The problem is that it assumes that the amount of the sale is independent of the probability of responding to the advertisement.<br />
<br />
If this assumption doesn't make sense then you might consider a "Heckit" (Heckman) model. It's like the two stage approach described by Yi-Chun but when you build the regression model you include a term (the "Inverse Mills Ratio") that comes from the Logistic (technically, a 'Probit')…
Yi-Chun's response is a fairly common approach (and a good description).<br />
<br />
The problem is that it assumes that the amount of the sale is independent of the probability of responding to the advertisement.<br />
<br />
If this assumption doesn't make sense then you might consider a "Heckit" (Heckman) model. It's like the two stage approach described by Yi-Chun but when you build the regression model you include a term (the "Inverse Mills Ratio") that comes from the Logistic (technically, a 'Probit') model.<br />
<br />
The method is described in detail in Madalla's (excellent) "Limited Dependent Variables" book, but it's also covered in Greene's "Econometrics" (and there's almost always a copy of Greene in someone's office/cubicle).<br />
<br />
Note: I prefer the "Heckit" to the "Tobit", particularly when I believe (and I think this is your case) that the sales amount is NEGATIVELY correlated with the response. Weight the solution by the sa…tag:www.analyticbridge.datasciencecentral.com,2009-07-23:2004291:Comment:486812009-07-23T00:29:11.938ZDaniel Kocishttps://www.analyticbridge.datasciencecentral.com/profile/DanielKocis
Weight the solution by the sales amount
Weight the solution by the sales amount Hi,
My SAS instructor with wh…tag:www.analyticbridge.datasciencecentral.com,2009-07-22:2004291:Comment:486742009-07-22T20:05:39.437ZYi-Chun Tsaihttps://www.analyticbridge.datasciencecentral.com/profile/YiChunTsai
Hi,<br />
My SAS instructor with whom I took course gave me an idea on how to do it. Basically, you have to build two models. One logistic regression model that predicts the probability of each customer respohnding to your advertisement, and the second linear regression model that predicts the sales amount of each customer. The trick is that when you build logistic regression model, you have to include every customer regardless of whether he/she responded or not whereas you just have to include the…
Hi,<br />
My SAS instructor with whom I took course gave me an idea on how to do it. Basically, you have to build two models. One logistic regression model that predicts the probability of each customer respohnding to your advertisement, and the second linear regression model that predicts the sales amount of each customer. The trick is that when you build logistic regression model, you have to include every customer regardless of whether he/she responded or not whereas you just have to include the customers who actually responded to your campaign with positive sales amount. After that, you just multiply the predicted probability with the predicted sales amount and rank/score your customers with the product of the two quantities.