Managing the Supply Chain

FORECASTING

Purspose is to explain as much of the systematic variability as possible and describe or quantify the unsystematic variability to support a decision. Is there a trend? How much variability? Forecast is a range not a number. Long term forecasts are less accurate. Aggregate forecasts are more accurate, greater forecasts errors for more upstream companies: bullwhip effect.

Measure of forecast error: e= F-D, we also have absolute forecast error ( the same but with absolute values)

The absolute percentage error: absolute error divided by D IeI/D

Measures of accuracy: MAD, MSE, MAPE ( Mean absolute Percentage Error) Avg of absolute percentage errors. Is a relative measure of forecast accuracy. Fair comparison between large demand and small demand

Accuracy 1-MAPE

Measure of bias: MFE , TS, RSFE. Forecast is biased when It constantly over or under estimates demand. Positive or negative values of MFE and RSFE tell us if the forecast is pessimistic or optimistic

TS: (-0.5,0.5) si esta fuera de este rango entonces el forecast is biased

Level is the avg around which observations vary

Trend is a predictable increase o decrease in the level over time. In the time series with trend the level is constantly shifting

Seasonality is a pattern of predictable and recurring shifts in the level

Random noise are unpredictable variations in the demand pattern

A time series with no systematic variability is also called stationary time series, the forecast for all future period is the same.

No systematic variability moving avg and exponential smoothing can be used.

With exponential smoothing varing the smoothing parameter places automatically more or less weight on old data

Trend but not seasonality> holts method > this forecasts are more accurate and less biased than simple exponential smoothing forecasts

INVENTORY

1) Economies of scale

2) Uncertainties (in demand, lead time, supply, etc.)

3) Speculation

4)...