L15 Forecasting

What is forecasting?

A technique to predict future values of a time series using past data and probability.

What are major requirements of a forecasting method?

Accuracy, minimal cost, simplicity, stability, and responsiveness.

Why is demand forecasting important in manufacturing?

Helps with capacity planning, staffing, training, finance, and stock availability.

What types of historical data are used in forecasting?

Direct data (e.g., sales figures) and indirect data (e.g., strikes, weather, legislation).

What are key forecasting principles?

Use similar time periods, enough data, and test multiple methods.

What is the time series approach?

Uses a variable's own past values to predict future values.

What is the econometric approach?

Uses relationships with other variables to make predictions.

What are examples of time series data?

Product demand, costs, market share, interest rates, and consumer trends.

What are the four typical patterns in time series data?

Trend, Cyclical, Seasonal, and Random.

What are the weaknesses of the time series method?

Irregularity and seasonality can obscure trends; no error margin estimation.

What is the moving averages method?

Forecasting using averages of past data over a chosen number of periods.

What are limitations of the moving averages method?

Subjective, sensitive to chosen period, equal weighting of all periods.

What is exponential smoothing?

A weighted moving average that gives more weight to recent data points.

What is the formula for exponential smoothing?

Ft = Ft-1 + α(Aₜ₋₁ - Ft-1)

What do the symbols in the exponential smoothing formula represent?

Ft = forecast, Ft-1 = previous forecast, Aₜ₋₁ = previous actual, α = smoothing factor.

What is the effect of a low α in exponential smoothing?

Smooths fluctuations, less responsive.

What is the effect of a high α in exponential smoothing?

More responsive to recent changes.

What range is typical for α in exponential smoothing?

Between 0.05 and 0.3.

What is econometric modelling?

Forecasting based on relationships with other variables using statistical models.

Give an example of an econometric model.

Battery sales = C₀ + C₁(car sales) + C₂(truck sales) + error term.

What are positive and negative correlations in econometrics?

Positive: variables move together; Negative: one increases while the other decreases.

What are limitations of forecasting?

Accuracy is often low; time series best for short-term, econometrics for long-term.

What is cusum in forecasting?

Cumulative sum of forecast errors used to monitor bias.

How is standard deviation used in forecasting?

Measures forecast error; helps create confidence intervals (e.g., ±2 SD = 95% confidence).

What is the goal of monitoring forecasting accuracy?

To detect bias and improve reliability of forecasts.