0

I have a dataset with 100+ inputs, and 6 inputs can be combined into a set, they collected from the same sensor. I same trying to find which sensor's data are the best features. Can Featuretools do feature engineering and find the best set of features from sets of features?

drerD
  • 629
  • 1
  • 9
  • 24

1 Answers1

1

This sounds more like a feature selection than feature engineering problem. Feature selection is the process of determine which features to use, while feature engineering is about create new predictive features. You can read more in this blog post.

You could use Featuretools if you want to create new features, but it won't directly tell you which sensor is best.

If you update your answer with an example dataset and expected output, we may be able to help more.

Max Kanter
  • 2,006
  • 6
  • 16
  • Thanks, my dataset is just a csv file of 7000 row and 120 columns of floating point values that I collected from sensors, and I'm to classify each row, which represent a time point, into three classes. I think I am a step behind from using Featuretools. – drerD Oct 29 '18 at 23:15