scripts/process-data.py

"""
# download parquet file from here: https://huggingface.co/datasets/Sachinkelenjaguri/Resume_dataset
# asked chatgpt to write me a script to convert and deduplicate
"""
import pandas as pd

# Step 1: Read the parquet file
df = pd.read_parquet('/Users/vincent/Downloads/csv-train.parquet')

if 'Category' in df.columns:
    unique_classes = df['Category'].unique()
    print("Unique classes in 'Category' column:")
    for cls in unique_classes:
        print(cls)
else:
    print("'Category' column does not exist in the data.")

# Step 2: Check if 'Resume' column exists
if 'Resume' in df.columns:
    # Keep only the 'Resume' column
    print(df.shape)
    df = df.drop_duplicates(subset=['Resume'])
    print(df.shape)
    df = df[['Resume']]
    # Remove all the new lines from each cell of the 'Resume' column
    df['Resume'] = df['Resume'].replace('\n',' ', regex=True)
else:
    print("'Resume' column does not exist in the data.")

# Step 3: Write the filtered dataframe back to a csv file
df.to_csv('/Users/vincent/Downloads/output.csv', index=False, header=False)

print("Completed successfully")