Learning Objectives

By the end of this lesson, learners should be able to:

  • Calculate forecast errors.
  • Use MAE, MSE, RMSE, and MAPE.
  • Compare forecasting models.
  • Interpret forecast results for managerial decisions.

Forecast Error

Forecast Error = Actual − Forecast

Example:

Actual sales = 120
Forecast sales = 110

Error = 10

Positive error means the forecast was too low.

Mean Absolute Error (MAE)

Average of absolute errors.

Easy to interpret because it uses original units.

Mean Squared Error (MSE)

Average of squared errors.

Penalizes large errors more heavily.

Root Mean Squared Error (RMSE)

Square root of MSE.

Expressed in original units and widely used.

Mean Absolute Percentage Error (MAPE)

MAPE = Average absolute percentage error.

Example: MAPE of 8% means forecasts are off by about 8% on average.

Comparing Models

Model

MAPE

Naïve

15%

Moving Average

10%

Exponential Smoothing

8%

Exponential smoothing is preferred because it has the lowest MAPE.

Business Interpretation

Forecast accuracy should be evaluated relative to business impact.

Example

  • Forecast error of 5 units may be acceptable for low-cost products.
  • The same error may be unacceptable for expensive aircraft engines.

Forecast Bias

Consistent over-forecasting or under-forecasting indicates bias.

Bias can lead to:

  • Excess inventory,
  • Stockouts,
  • Staffing inefficiency,
  • Cash flow problems.

Monitoring Forecast Performance

Organizations should regularly:

  • Compare forecasts with actuals,
  • Track error metrics,
  • Re-estimate models,
  • Investigate major deviations.

International Case Study

A global apparel company forecasted winter jacket demand. Initial MAPE was 18%. After incorporating weather data and regional seasonality, MAPE fell to 9%, reducing excess inventory by millions of dollars.

Communicating Forecasts

Reports should include:

  • Forecast value,
  • Confidence range,
  • Assumptions,
  • Key risks,
  • Recommended actions.

Example

“Next quarter revenue is forecast at USD 24 million ± USD 2 million, assuming current exchange rates and marketing expenditure remain unchanged.”

Practical Activity

Calculate MAE and MAPE for three forecasting methods using a small sales dataset and recommend the best method.

Learning Materials / Reference Materials

  • Hyndman & Athanasopoulos. Forecasting: Principles and Practice.
  • Makridakis Forecasting Competition Resources.

Lesson Summary

Forecast evaluation is essential because predictive models must be measured against actual outcomes. Accuracy metrics help organizations select better models and improve business planning.

Lesson Quiz (5 MCQs)

Question 1 Forecast error equals:

  1. Actual − Forecast
    B. Forecast − Actual always
    C. Actual + Forecast
    D. Forecast ÷ Actual

Question 2 MAE uses:

  1. Absolute errors
    B. Squared errors only
    C. Percentages only
    D. Maximum errors only

Question 3 RMSE is expressed in:

  1. Original data units
    B. Percent only
    C. Probability units only
    D. Currency only

Question 4 Lower MAPE indicates:

  1. Better forecast accuracy
    B. Worse forecast accuracy
    C. More data points only
    D. Higher revenue automatically

Question 5 Consistent over-forecasting indicates:

  1. Forecast bias
    B. Perfect accuracy
    C. Seasonality only
    D. Random variation only

Correct Answers

  1. A
  2. A
  3. A
  4. A
  5. A

Topic Conclusion

Predictive analytics and forecasting enable organizations to move from reactive management to proactive planning. By understanding time-series behavior, selecting appropriate forecasting methods, applying regression models, and evaluating forecast accuracy, managers can make better decisions regarding inventory, staffing, budgeting, marketing, investment, and operations. Effective forecasting does not eliminate uncertainty, but it reduces uncertainty sufficiently to improve organizational performance and strategic agility.

Topic Practical Assignment

A multinational retail company operating in New York, London, Berlin, Dubai, Singapore, and Sydney wants to forecast monthly sales for the next six months.

Using historical monthly sales data:

  1. Plot the time series.
  2. Identify trend and seasonality.
  3. Calculate forecasts using:
    • Naïve method,
    • 3-month moving average,
    • Weighted moving average,
    • Exponential smoothing.
  4. Compute MAE and MAPE for each method.
  5. Select the best forecasting model.
  6. Prepare a sales forecast dashboard.
  7. Write a management report explaining:
    • Forecast assumptions,
    • Expected sales,
    • Risks,
    • Recommended inventory and staffing actions.

The report should be suitable for presentation to senior management.

Previous