Unverified Commit 0f09420d authored by Claude's avatar Claude
Browse files

Fix MAPE calculation to use correct denominator 

- Changed calculate_mape to divide by valid_count (non-zero actuals) instead of total count
- Added proper validation checks for empty/mismatched arrays
- Made accuracy calculation more explicit with clear loop
- Fixed bug where MAPE was incorrectly calculated when some actual values were zero
parent 4ffedd7c

forecast_app/app.py

+21 −8
Original line number Diff line number Diff line
@@ -44,6 +44,7 @@ def calculate_mape(actual_values, predicted_values): 
    Calculate Mean Absolute Percentage Error (MAPE)



    MAPE = (1/n) * Σ(|actual - predicted| / |actual|) * 100

    where n is the count of non-zero actual values



    Args:

        actual_values: List of actual values
@@ -52,17 +53,25 @@ def calculate_mape(actual_values, predicted_values): 
    Returns:

        MAPE as percentage

    """

    if len(actual_values) == 0 or len(predicted_values) == 0:

    if not actual_values or not predicted_values:

        return 0.0



    if len(actual_values) != len(predicted_values):

        return 0.0



    n = len(actual_values)

    total_percentage_error = 0.0

    valid_count = 0



    for actual, predicted in zip(actual_values, predicted_values):

        if actual != 0:

            total_percentage_error += abs(actual - predicted) / abs(actual)

        if actual != 0:  # Only include non-zero actuals

            percentage_error = abs(actual - predicted) / abs(actual)

            total_percentage_error += percentage_error

            valid_count += 1



    if valid_count == 0:

        return 0.0



    mape = (total_percentage_error / n) * 100

    mape = (total_percentage_error / valid_count) * 100

    return mape
@@ -79,15 +88,19 @@ def calculate_accuracy(actual_values, predicted_values): 
    Returns:

        Accuracy as percentage

    """

    if len(actual_values) == 0 or len(predicted_values) == 0:

    if not actual_values or not predicted_values:

        return 0.0



    if len(actual_values) != len(predicted_values):

        return 0.0



    total_actual = sum(actual_values)

    if total_actual == 0:

        return 0.0



    total_absolute_error = sum(abs(actual - predicted)

                               for actual, predicted in zip(actual_values, predicted_values))

    total_absolute_error = 0.0

    for actual, predicted in zip(actual_values, predicted_values):

        total_absolute_error += abs(actual - predicted)



    accuracy = (1 - (total_absolute_error / total_actual)) * 100

    return accuracy