Forecasting Techniques

Why it matters

Forecasting techniques are crucial in industrial engineering as they help predict future trends, enabling effective decision-making and resource planning. Accurate forecasts can lead to optimized production schedules, inventory management, and improved customer satisfaction.

Key ideas

• Forecasting: The process of predicting future events based on historical data and analysis.
• Qualitative Methods: These include expert judgment, market research, and Delphi method, often used when historical data is unavailable.
• Quantitative Methods: These rely on numerical data and include time series analysis, causal models, and econometric models.
• Time Series Analysis: Involves methods like moving averages, exponential smoothing, and ARIMA models to analyze data points collected or recorded at specific time intervals.
• Causal Models: These models assume that the variable to be forecasted has a cause-and-effect relationship with one or more other variables.

Formulas

• Simple Moving Average: SMA = (D1 + D2 + ... + Dn) / n

  • SMA: Simple Moving Average
  • D1, D2, ..., Dn: Demand in each period
  • n: Number of periods

• Exponential Smoothing: Ft = α * Dt-1 + (1 - α) * Ft-1

  • Ft: Forecast for the current period
  • α: Smoothing constant (0 < α < 1)
  • Dt-1: Actual demand in the previous period
  • Ft-1: Forecast for the previous period

Simple exponential smoothing models a level without explicit trend or seasonality. A larger α reacts faster but can follow noise. Evaluate forecasts on later held-out periods rather than fitting and judging on the same observations. Regression association alone does not establish causation; check whether predictors will be available when the forecast is made. Forecast errors and uncertainty matter alongside the point forecast.

Worked example

Given: A company records the demand for a product over the last four months as 100, 120, 130, and 110 units. Calculate the forecast for the next month using a 3-month simple moving average.

1. Identify the periods: Last three months' demand = 120, 130, 110 units.
2. Apply the formula: SMA = (120 + 130 + 110) / 3
3. Calculate: SMA = 360 / 3 = 120

Forecast for the next month: 120 units

Common mistakes

• Confusing qualitative and quantitative methods.
• Incorrectly applying the smoothing constant in exponential smoothing.
• Using outdated or irrelevant data for forecasting.

For GATE ME

Questions often involve calculating forecasts using time series methods like moving averages or exponential smoothing. Practice problems that require understanding the assumptions and limitations of different forecasting techniques.

Quick check

1. What is the main difference between qualitative and quantitative forecasting methods?
2. How does exponential smoothing differ from simple moving average?
3. Why is it important to choose the correct smoothing constant in exponential smoothing?

Answers: 1. Qualitative methods rely on judgment and opinion, while quantitative methods use numerical data. 2. Exponential smoothing gives more weight to recent observations. 3. The smoothing constant affects the sensitivity of the forecast to changes in data.