1

I have data as follows in order to do a predictive learning as to what feature do people find attractive in a model when purchasing clothes online.

So I have data as follows.

COLORofCLOTHING  MODELHAIR_COLOR MODEL_BUILD SELLER_CATEGORY
  Red               Black         Lean           1
  Blue              Brown         Lean           5
  Black             Blonde        Healthy        10

In order to predict if the clothing will sell well given a set of attributes. However seller category can be anything between 1 to 10 (1 being best and 10 being worst) I am not sure how to approach this problem. I am using weka for this purpose. Can people please give me ideas on how to approach this problem?

basically I want to build a model which learns the features like color of the clothing etc and can predict how well the clothes will sell.

ExceptionHandler
  • 213
  • 1
  • 8
  • 24

1 Answers1

2

Transform and normalise your dataset into something along the lines of:

 color_red   color_blue color_black  hair_black  hair_brown  hair_blonde ... prediction
 1           0          0            1           0           0           ... 0
 0           1          0            0           1           0           ... 0.5
 0           0          1            0           0           1           ... 1

Random Forests and Neural Networks should be able to give you predictions.

ilikedata
  • 121
  • 2
  • Thanks for the tip. I am familiar with algorithms like JRip, Bayes, M5P, J48, LWL. Do you think there is a way of doing this using the known algorithms? (Just for my comfort) – ExceptionHandler Dec 21 '12 at 04:44
  • I'd be surprised if you couldn't get results. You may find they give good results or perhaps not. You'll find this out through experimentation. – ilikedata Dec 21 '12 at 05:07
  • I tried to do it as a classification problem but it gave terrible results. (100% error rate). I might be doing something wrong there. I want to do it as a regression problem but unfortunately I am doing something wrong with regression. Is there something in specific I need to take care off in order to work on this as regression problem? – ExceptionHandler Dec 21 '12 at 05:18
  • With your data transformed as above you should be able to treat it as a regression problem. – ilikedata Dec 21 '12 at 12:42