debug_csv.py

import pandas as pd
import os

# Check the CSV file
csv_path = 'data/Consumer_Complaints_train.csv'
print(f"Checking CSV file: {csv_path}")
print(f"File exists: {os.path.exists(csv_path)}")

# Load the CSV
df = pd.read_csv(csv_path)
print(f"Columns: {df.columns.tolist()}")
print(f"Shape: {df.shape}")
print("Data:")
print(df.to_string())

# Check for empty narratives
print("\nChecking for empty narratives:")
print(f"Rows with empty narrative: {df['Consumer complaint narrative'].isna().sum()}")
print(f"Rows with empty string narrative: {(df['Consumer complaint narrative'] == '').sum()}")

# Check narrative lengths
print("\nNarrative lengths:")
print(df['Consumer complaint narrative'].str.len())

# Check if narratives are longer than 50 characters
print("\nNarratives longer than 50 chars:")
print((df['Consumer complaint narrative'].str.len() > 50).sum())