|
|
- Features in Wellness:
- Pain [0, 1] - no NaNs
- Illness [0, 0.5, 1] - no NaNs
- Menstruation [0, 1] - 16 NaNs, filled with 0. Not a big statistical difference, so this is fine
- Nutrition [0, 0.5, 1] - 837 NaN, filled with 0. Not a useful feature
- NutritionAdj [0, 1] - 745 NaN, filled with 0. Again not useful
- USGMeasurement [0, 1] 168 NaN, filled with 0.
- USG [1.0...] 4382 NaN, not a useful feature
- TrainingReadiness [0..1] - no NaNs
-
- Useful features include Pain, Illness, Menstruation, TrainingReadiness
- The others either have too many NaNs present to extract any useful meaning or are just unhelpful features
- to begin with, like Nutrition.
-
-
- Notnormalized_with_0NaN_wellness.csv:
-
- - The only feature of significance that had NaN values put into it were Menstruation, as only 16 NaNs were present
- and wouldn't present any statistical difference either way.
-
- - Working in the notnormalized_with_0NaN_wellness csv should be functional, just have to remove any string columns
- before putting into algorithms as they are not removed in this version
|
