# Get best performing feature

A data set that looks like this

Person Money     b     c
1        200    83     6
1         70    55     2
1          5    27    NA
2         40    27     1
2        304     7    NA
2         20     3     3
2         60     5     4


From every column I want to extract dynamic features with thresholds. For example, for Money 'the amount of times Money of person ... is < 100'. However, I'm not sure which threshold will give the best predictive power.

• the amount of times Money of person ... is < 100
• the amount of times Money of person ... is < 90
• the amount of times Money of person ... is < 80 etc

So I would like to figure out which threshold is the best choice. I'm really new to ML and I have difficulties thinking of the best way to approach this. I thought about:

• Extract desired feature with all possible thresholds using feature selection method (wrapper, filter)
• Manually test all thresholds and look which threshold gives highest accuracy

Any suggestions would be appreciated!