| ''' | |
| import pandas as pd | |
| from sklearn.utils import resample | |
| # Load the cleaned dataset | |
| df = pd.read_csv("cleaned_qbias_dataset.csv") | |
| # Separate majority and minority classes | |
| df_left = df[df.label == 0] | |
| df_center = df[df.label == 1] | |
| df_right = df[df.label == 2] | |
| # Determine target size (matching the majority class: 'left') | |
| target_size = len(df_left) | |
| # Upsample center class to match 'left' | |
| df_center_upsampled = resample( | |
| df_center, | |
| replace=True, | |
| n_samples=target_size, | |
| random_state=42 | |
| ) | |
| # Combine all classes into one balanced DataFrame | |
| df_balanced = pd.concat([df_left, df_center_upsampled, df_right]) | |
| # Shuffle the final dataset | |
| df_balanced = df_balanced.sample(frac=1, random_state=42).reset_index(drop=True) | |
| # Save to new CSV | |
| df_balanced.to_csv("cleaned_qbias_balanced.csv", index=False) | |
| print("Balanced dataset saved as cleaned_qbias_balanced.csv") | |
| print(df_balanced['label'].value_counts()) | |
| ''' | |
| import pandas as pd | |
| df = pd.read_csv("cleaned_qbias_dataset.csv") | |
| print(df['label'].value_counts()) |