About EricTsai

EricTsai · ‎05-17-2018

Hi, I am building a predictive response model using proc logistic. However, due to nonlinearity between predictors and the response, I tried to use PROC ADAPTIVEREG for regression spline basis functions to accommodate for non-linearity. I also tried PROC GAMPL to get splines but I didn't know how to get the scoring code that will do the transformation from the procedure's output. I want to use the spline transformation to score my future new data and need the scoring code with transformation.

EricTsai · ‎01-27-2016

Hi, I have Enterprise Miner 13.2 but I don't have Credit Scoring Node to helpe me with optimal binning of my continuous, ordinal and nominal predictive variables. Does anyone have any suggestions on how to do it without Credit Score Node? Thanks

EricTsai · ‎10-21-2015

Hi, Miguel: Each campaign's target is a binary variable, i.e., Y=1 if customer click and Y=0 if customer didn't click on the web link. I actually built a decision tree with nominal target variable exactly by doing what you suggested. However, my concern is that not all customers were exposed to all my 4 campaigns. In other words, it didn't really make sense to say that customer John prefered Campaign A over Campaign B, because John was only exposed to Campaign A, but not Campaign B.

EricTsai · ‎10-19-2015

Hi, I have multiple campaigns and their performance data. I built one decision tree for each campaign to determine the most important independent variables and the rules that can be implemented for each campaign. Each tree gives me leaves/ segments that tells me the group of customers least or most likely to respond to each campaign. However, it turns out that my clients want to have just one tree to do the similar job. However, my trees look so different from each other that I don't know how to build just one tree to do it. Any ideas?

EricTsai · ‎07-17-2015

Hi, I built several incremental response models using both SAS Enterprise Miner and SAS/STAT. The paper that Miguel cited was very helpful to me when I used Enterprise Miner to build models. Also, I would recommend you to take the SAS training course in this topic. The course covers more techniques using SAS/STAT. Hope this helps a bit.

EricTsai · ‎07-14-2015

Hi, Any idea on how to build a incremental sales model for marketing campaign? I tried it before but the performance is not very stable. Sometimes, it ranks but other times, I would see big incremental sales coming from bottom decile. Please help.

EricTsai · ‎07-01-2015

Hi, I have heard many success story using SVM in predictive modeling but according to my own experience, my SVM didn't outperform my basic Logistic regression in Enterprise Miner. Could someone help me in fine tuning my SVM? I believe I haven't done a great job in fine tuning my SVM in EMiner and that caused the underperformance of my SVM. Any papers or any tutorials are more than welcome!

EricTsai · ‎06-26-2015

Hi, Miguel: In your experience, if you have thousands of potential independent variables to start with in your modeling process, what would you try to reduce this huge number to a more manageable list? Have you seen any method that give you a more predictive model at the end? What is the state of art in this area right now at SAS?

EricTsai · ‎06-26-2015

Hi, Does anyone know how to use random forest in Enterprise Miner to do variable selection? I have large number of original variables to start with ranging from 7,000 to 8,000. I have tried using Variable Clustering Node in EMiner but would love to learn more ways of variable selection for large data set. Thanks a lot.

EricTsai · ‎06-12-2015

Thank you very much Patrick for sharing this information and specifically your code. It really helps in explaining the difference between the output from PROC DISCRIM and MBR node.

EricTsai · ‎06-10-2015

Hi, Reeza: I used PROC DISCRIM with METHOD=NPAR, which in terms gives me KNN (k-nearest neighbors) algorithm. KNN is NOT an unsupervised learning algorithm; it is a supervised learning algorithm.

EricTsai · ‎06-08-2015

Hi, Miguel: Do you know the difference between PROC DISCRIM and MBR node in terms of KNN? I used both but got a totally different result. I don't know which one to use for scoring my new data now.

EricTsai · ‎06-08-2015

Hi, Ralph: Do you know the difference between MBR node and PROC DISCRIM procedure? I used both to run my KNN but I got totally different result, but I don't know which one to use for scoring my new data set. Thanks for your help.

EricTsai · ‎10-06-2014

This example is very good as introduction to this node. I tried using this node following the paper but when I tried to valided the ranking using off-time sample, it didn't work any more. In other words, when I used campaign data that is different from the campaign for model building, the result totally messed up. The model is quite unstable. Is there anythomg else that I can try to correct it? Thanks.

EricTsai · ‎07-14-2010

Hi, all: I have set of explanatory variables with hig collinearity to build my time series regression model. One of my friend used ridge regression to deal with this problem in time series data. I never heard of people using ridge regresssion in time series modeling and I am very doubtful of his approach. Could you share with me the ways to deal with high collinearity among explanatory variables in time series forecasting models? Thanks.

Online Status	Offline
Date Last Visited	‎06-07-2019 09:13 PM

How to get scoring function from PROC GAMPL

How to decide on binning when calculating WOE

Re: Multiple Campaigns with Multiple Decision Trees

Multiple Campaigns with Multiple Decision Trees

Re: Uplift Modelling, Please Help..Thank You

Incremental Sales Model

Fine tune my SVM in EM

Re: Random Forest for Variable Selection

Random Forest for Variable Selection

Re: RD-Tree algorithm in MBR node

Re: Optimization Problem

Re: RD-Tree algorithm in MBR node

view enteprise miner output in enterprise guide

Re: Tip: How to Select the Optimal Targets in Direct Marketing Using S...

How to get scoring function from PROC GAMPL

How to decide on binning when calculating WOE

Re: Multiple Campaigns with Multiple Decision Trees

Multiple Campaigns with Multiple Decision Trees

Re: Uplift Modelling, Please Help..Thank You

Incremental Sales Model

Fine tune my SVM in EM

Re: Random Forest for Variable Selection

Random Forest for Variable Selection

Re: RD-Tree algorithm in MBR node

Re: Classification: K nearest neighbors (MBR)

Re: Classification: K nearest neighbors (MBR)

Re: RD-Tree algorithm in MBR node

Re: Tip: How to Select the Optimal Targets in Direct Marketing Using S...

high collinearity among explanatory variables in forecasting models