TEST1

• Preprocessing none

• Prediction

  RandomForestClassifier

  training_data.csv (training) testing_data.csv (for prediction) ⇒ predict.py ⇒ predicted_results.csv

TEST2

• Preprocessing

  1. replacing missing value with mean

     training_data.csv ⇒ impute.py ⇒ data_filled_mean.csv

  2. resample data with SMOTE

     data_filled_mean.csv ⇒ resample.csv ⇒ data_resampled.csv

• Prediction

  scaling then use tf.keras.models.Sequential for training/prediction

  data_resampled.csv (training) testing_data.csv (for prediction) ⇒ predict.py ⇒ predicted_results.csv

TEST3

• fill the first col-name with id in both testing/training data

• Preprocessing

  1. replacing missing value with mean

     training_data.csv ⇒ impute.py ⇒ data_filled_mean.csv

  2. resample data with SMOTE

     data_filled_mean.csv ⇒ resample.csv ⇒ data_resampled.csv

• Prediction

  scaling then use tf.keras.models.Sequential for training/prediction

  data_resampled.csv (training) testing_data.csv (for prediction) ⇒ predict.py ⇒ predicted_results.csv