python - divide into test set and training set with all rows corresponding to one combination of attribute values either in the training or the test set

Question

My input file is under the form:

gold,ProgramName,RequirementID,MethodID,DataTypeName,DataTypeID,FieldMethodOwnerClass,VariableName,fieldMethodID
Trace,chess,1,1,boolean,0,1,_moveRight,3
Trace,chess,1,1,boolean,0,1,_computerIsWhite,4
NoTrace,chess,2,1,boolean,0,1,_moveRight,3
NoTrace,chess,2,1,boolean,0,1,_computerIsWhite,4
NoTrace,chess,3,1,boolean,0,1,_moveRight,3
NoTrace,chess,3,1,boolean,0,1,_computerIsWhite,4
NoTrace,chess,4,1,boolean,0,1,_moveRight,3
NoTrace,chess,4,1,boolean,0,1,_computerIsWhite,4
NoTrace,chess,5,1,boolean,0,1,_moveRight,3
NoTrace,chess,5,1,boolean,0,1,_computerIsWhite,4
NoTrace,chess,6,1,boolean,0,1,_moveRight,3
NoTrace,chess,6,1,boolean,0,1,_computerIsWhite,4
NoTrace,chess,7,1,boolean,0,1,_moveRight,3
NoTrace,chess,7,1,boolean,0,1,_computerIsWhite,4
NoTrace,chess,8,1,boolean,0,1,_moveRight,3
NoTrace,chess,8,1,boolean,0,1,_computerIsWhite,4
NoTrace,chess,1,4,de.java_chess.javaChess.game.Game,50,1,_game,1
NoTrace,chess,1,4,byte,0,67,KING,353
NoTrace,chess,1,4,byte,0,67,PAWN,348
NoTrace,chess,1,4,de.java_chess.javaChess.game.Game,50,1,_game,1
NoTrace,chess,2,4,de.java_chess.javaChess.game.Game,50,1,_game,1
NoTrace,chess,2,4,byte,0,67,KING,353
NoTrace,chess,2,4,byte,0,67,PAWN,348
NoTrace,chess,2,4,de.java_chess.javaChess.game.Game,50,1,_game,1
NoTrace,chess,3,4,de.java_chess.javaChess.game.Game,50,1,_game,1
NoTrace,chess,3,4,byte,0,67,KING,353
NoTrace,chess,3,4,byte,0,67,PAWN,348

I would like to have my data split into a training and a test set but I would like to have all rows corresponding to 1 combination of <requirementID, MethodID> into the training set or into the test set. For instance, you notice that the 2 first rows both have <1,1> as values for the <requirementID, MethodID> tuple. The situation I want to avoid is the following: I don't want to have the first row in the training set and the second row in the test set. I would like to have both of them either in the training set or in the test set. Thus, all rows correponding to one combination of <requirementID, MethodID> should be either in the training set or the test set. For instance, the two first rows with <1,1> as values for <RequirementID,MethodID> should be in the training set and the third and fourth rows with values <2,1> should be in the test set.

Here is the Python code I am using. As you notice, it's splitting randomly into a test and a training set, which I don't want, I want the plit to be made according to the rules described above:

import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import classification_report, confusion_matrix, accuracy_score
import sys


def main():
  
    dataset = pd.read_csv( 'inputFields.txt', sep= ',', index_col=False) 

    #convert Inner, Root, Leaf into 0, 1, 2
   
    dataset['ProgramName'] = dataset['ProgramName'].astype('category').cat.codes
    dataset['DataTypeName'] = dataset['DataTypeName'].astype('category').cat.codes
    dataset['VariableName'] = dataset['VariableName'].astype('category').cat.codes
    dataset['gold'] = dataset['gold'].astype('category').cat.codes
    
    pd.set_option('display.max_columns', None)
    
    row_count, column_count = dataset.shape
     
        
    X = dataset.iloc[:, 1:column_count].values
    y = dataset.iloc[:, 0].values

    print(y)

    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.4, random_state=1)    
    
         
    ################################################################################
       
        
        
    classifier = RandomForestClassifier(n_estimators=400, random_state=0)
    classifier.fit(X_train, y_train)
    y_pred = classifier.predict(X_test)
        
    
        
    print('confusion matrix
',confusion_matrix(y_test,y_pred))
    print('classification report
', classification_report(y_test,y_pred))
    print('accuracy score', accuracy_score(y_test, y_pred))     
if __name__=="__main__": 
    
        main()

question from:https://stackoverflow.com/questions/65925829/divide-into-test-set-and-training-set-with-all-rows-corresponding-to-one-combina

Answer 1

Waitting for answers