"No one is harder on a talented person than the person themselves" - Linda Wilkinson ; "Trust your guts and don't follow the herd" ; "Validate direction not destination" ;

October 27, 2017

Day #75 - Missing Values

  • Reasons for Missing Values
  • How to Engineer them effectively
  • Hidden Missing Values
  • Plot distribution of values and find from histogram
Filling missing Values
  • -999, -1 (Fill with some value) - useful to provide different category, Perf Suffers
  • mean, median
  • Reconstruct value
  • add isnull column
Reconstruction
  • Missing values in timeseries
  • Temperature values missing for some days of month
  • Based on increase / decrease pattern
  • Ignore missing value while calculating mean
  • Change Categories to frequencies
  • XGBoost can handle NAN
Happy Learning and Coding!!!

No comments: