src/analysis.py

# src/analysis.py
import pandas as pd

def calculate_correlations(df: pd.DataFrame) -> dict:
    correlations = {}
    # Overall correlation
    overall_corr = df[['Feedback_Stars', 'Instructor_Rating']].corr().iloc[0, 1]
    correlations['Overall'] = overall_corr

    # Per subject correlation
    for subject in df['Subject'].unique():
        subject_df = df[df['Subject'] == subject]
        if len(subject_df) > 1: # Need at least 2 data points for correlation
            subject_corr = subject_df[['Feedback_Stars', 'Instructor_Rating']].corr().iloc[0, 1]
            correlations[subject] = subject_corr
        else:
            correlations[subject] = None # Not enough data
    print("Correlation analysis complete.")
    return correlations