helper.py

"""
Module for generating course recommendations based on similarity scores.
"""

def normalize_rating(rating_str):
    """
    Normalize the course rating to a 0-1 scale.
    
    Converts the rating from a 0-5 scale to a 0-1 scale for consistency in ranking.
    If the rating is not a valid number, it returns 0.
    """
    try:
        return (float(rating_str) - 0) / (5 - 0)  # Normalize to 0-1
    except ValueError:
        return 0  # Return 0 if the rating is invalid

def get_recommendations(course_name, data, similarity_matrix, top_n=6,
                        rating_weight=0.05):
    """
    Get top N course recommendations based on similarity to the given course name.
    
    This function retrieves the best matching course from the dataset and ranks other 
    courses based on their similarity score. It also adjusts recommendations based on 
    course ratings, and returns the top N recommendations with a final score.
    
    Parameters:
    - course_name: Name of the course to find recommendations for.
    - data: The DataFrame containing course information.
    - similarity_matrix: Precomputed similarity matrix.
    - top_n: Number of top recommendations to return.
    - threshold: Minimum similarity score to consider a match.
    - rating_weight: Weight to apply to course ratings in the final recommendation score.
    """
    course_name = data[data['Course Name'] == course_name]  # Filter data for selected course
    course_idx = course_name.index[0]  # Get the index of the selected course
    similarity_scores = list(enumerate(similarity_matrix[course_idx]))  # Get similarity for courses

    recommendations = []
    for idx, similarity_score in sorted(similarity_scores, key=lambda x: x[1], reverse=True)[:top_n]:
        course_data = data.iloc[idx]  # Get course data for the current recommendation
        normalized_rating = normalize_rating(course_data.get('Course Rating', '0'))

        # Prepare recommendation dictionary with relevant course information
        recommendations.append({
            "course_name": course_data['Course Name'],
            "course_url": course_data.get('Course URL', ''),
            "rating": course_data['Course Rating'],
            "institution": course_data.get('University', 'Unknown'),
            "difficulty_level": course_data.get('Difficulty Level', 'Unknown'),
            "similarity": similarity_score,
            "final_score": similarity_score * (1 - rating_weight) + normalized_rating * rating_weight
        })

    # Return sorted recommendations based on the final score
    return sorted(recommendations, key=lambda x: x['final_score'], reverse=True)