Supply Chain Forecasting

• Forecasting predicts happenings in the future

Qualitative Forecasts

• Qualitative forecasts are subjective

Quantitative Forecasts

• Quantitative forecasts use mathematical formulas

Strategic Role of Forecasting in Supply Chain Management

• Accurate forecasting determines inventory levels in the supply chain
• Continuous replenishment involves suppliers and customers sharing continuously updated data and is typically managed by the supplier; it also reduces company inventory and speeds up customer delivery
• Continuous replenishment variations:
  • Quick response, which allow retailers to accommodate 'fads'
  • Just-in-time (JIT)
  • Vendor-managed inventory (VMI)
  • Stockless inventory
• These systems rely heavily on accurate short-term forecasts

Forecasting and Management

• Accurate customer demand forecasting is key to providing good quality service
• Successful strategic planning requires accurate forecasts of future products and markets

Components of Forecasting Demand

• Time frame
• Demand behavior
• Causes of behavior

Time frame categories

• Short-range forecasts typically encompass the immediate future, up to six months, and are used for detailed scheduling of goods and services
• Medium-range forecasts cover six months to two years and are often used for aggregate planning, including human resources, inventory, and technology
• Long-range forecasts typically cover a period longer than two years, and can extend up to 50 years, with five years being typical and are used to make capital investment decisions
• Demand Behavior*
• Trend: Gradual, long-term up or down movement of demand
• Random variations: Movements in demand with no pattern
• Cycle: Repetitive up-and-down movement in demand
• Seasonal Pattern: Repetitive up-and-down movement in demand occurring periodically

Forecasting Methods

• Time series: Use historical demand data to predict future demand
• Regression methods: Develop a mathematical relationship between demand and factors
• Qualitative: Use of management judgment, expertise, and opinion to predict future demand

Internal Qualified Forecasts

• Management
• Marketing
• Purchasing
• Engineering

Delphi Method

• Involves soliciting forecasts about technological advances from experts

Forecasting Process

• Identify the purpose of the forecast
• Collect the historical data
• Plot the data and Identify patterns
• Select the appropriate forecast model
• Develop/compute forecast for period of historical data
• Check forecast accuracy with one or more measures
• If accuracy of forecast is not acceptable, select a new forecast model or adjust parameters of existing model. If acceptable:
• Forecast over the planning horizon
• Adjust forecast based on qualitative information and insight to measure accuracy

Time Series

• Time is the independent variable
• Assumptions are made that what happened in the past will continue to occur in the future
• Relate to only one factor, time
• Includes Naive forecast, moving average, exponential smoothing and linear trend line

Naive Forecast

• Demand in current period is used as next period’s forecast

Simple Moving Average

• It uses average demand for a fixed sequence of periods and is good for stable demand with no pronounced behavioral patterns

Weighted Moving Average

• Weights are assigned to most recent data

Exponential Smoothing method: characteristics

• Averaging method
• Weights most recent data more strongly
• Reacts more to recent changes
• Widely used, accurate method
• Applied to single stream sales data

Smoothing Constant

• Smoothing constant: α

Exponential Smoothing Formula

• F(t+1) = αD(t) + (1 - α)F(t)
  • Where:
    • F(t+1) = forecast for next period
    • D(t) = actual demand for present period
    • F(t) = previously determined forecast for present period
    • α = weighting factor, smoothing constant

Effect of Smoothing Constant in Exponential Smoothing

• If α = 0.20, then F(t+1) = 0.20D(t) + 0.80F(t)
• If α = 0, then F(t+1) = 0D(t) + 1F(t) = F(t); meaning that the forecast does not reflect recent data
• If α = 1, then F(t+1) = 1D(t) + 0F(t) = D(t); meaning that the forecast is based only on the most recent data

Adjusted Exponential Smoothing

• AF(t+1) = F(t+1) + T(t+1)
  • Where:
    • T = an exponentially smoothed trend factor
    • T(t+1) = β(F(t+1) - F(t)) + (1 - β) T(t)
    • T(t) = the last period trend factor
    • β = a smoothing constant for trend (0 ≤ β ≤ 1)

Linear Trend Line: Formula

• y = a + bx
  • Where:
    • a = intercept
    • b = slope of the line
    • x = time period
    • y = forecast for demand for period x

Linear Trend Line: Formula for b

• b = (Σxy - nxy) / (Σx^2 - nx^2)

Linear Trend Line: Formula for a

• a = y - bx
  • Where
    • n = number of periods
    • x = average mean of x values
    • y = average mean of y values

Seasonal Adjustments

• Repetitive increase/decrease in demand
• Use seasonal factor to adjust forecast

Seasonal factor: Formula

• Seasonal factor = S(i) = D(i) / ΣD

Forecast Error

• Forecast error is the difference between forecast and actual demand

Mean Absolute Deviation (MAD): Definition

• MAD is the mean absolute deviation

Formula to calculate Mean Absolute Deviation

• MAD = Σ|D(t) - F(t)| / n
  • Where:
    • t = period number
    • D(t) = demand in period t
    • F(t) = forecast for period t
    • n = total number of periods
    • || = absolute value

Other Accuracy Measures

• Mean absolute Percent Deviation (MAPD): Formula*
• MAPD = ∑|Dt - Ft|/∑Dt
• Cumulative Error: Formula*
• E = ∑et
• Average Error: Formula*
• (E) = ∑et/n

Forecast control

• To monitor the forecast to see if it is biased high or low, tracking signals are used
• 1 MAD ≈ 0.8 б
• Control limits of 2 to 5 MADs are used most frequently
• Tracking signal equation*
• Tracking signal = Σ(D(t) - F(t)) / MAD = E / MAD

Statistical Control Charts

• σ can be used in calculation of statistical control limits for the forecast error
• Control limits are tyically set at +- 3σ

Statistical Control Charts: Formula

• σ = square root of [(Σ(D(t) - F(t))^2) / (n-1)]
  • Where all variables are forecast values

Mean squared error (MSE)

Mean squared error (MSE): Definition

• The are the Average of squared forecast error