You are on page 1of 20

Forecasting [ref.

Chopra & Meindl pages 68 to 75] Forecasting is a scientific method of determining demand in future Starting point for planning activities in logistics Distribution plan production scheduling Materials planning and inventory control
1/24/2014 1

Some characteristics of forecasts


Forecasts are almost always wrong

Forecasts are more accurate for groups or


families of items

motor cars and models


Aggregate forecasts are more accurate annual rainfall and daily rainfall
1/24/2014 2

Forecasts are more accurate for short periods


(tomorrow, next year)

Forecast should include an estimate of error


Forecasts are no substitutes for facts

1/24/2014

Components of forecast Past demand Planned advertising or marketing efforts Planned price discounts State of economy Competitors actions forecasters knowledge and judgment
1/24/2014 4

Major categories of forecasts (forecasting methods) Qualitative & quantitative forecasts Qualitative forecasting Forecast is based on personal judgment Subjective (opinion based) can be obtained in less time

1/24/2014

When facts are unavailable for other methods


Made for specific items based on aggregate

forecast for markets

1/24/2014

Some qualitative methods of forecasting


Market surveys potential customers opinions

Delphi method
Panel consensus

Life cycle analogy


Informed judgment sales force

1/24/2014

Quantitative forecasting Fact based, scientific models Causal-Correlating demand to specific causal factors in environment. Estimate these causal factors and forecast demand. Ambient temperature and coffee consumption! Monsoon and rice

production!

1/24/2014

Econometric models-statistical analysis of various sectors of economy Input-output models

Output of an industry is input to another industry Forecast for paint from demand for automobiles Generally used for project needs
1/24/2014 9

Simulation using computer simulation to simulate sectors of economy Time series 1. Regression analysis Statistical method Developing analytical relationship between two variables
1/24/2014 10

Using statistical tools on past data to identify


trend, under stable environmental situations and demand 2. Moving average method Simple moving average estimator decides the period over which average is taken. 3 months or so
1/24/2014 11

MONTHS JANUARY FEBRUARY MARCH APRIL

ACTUAL 4200 4300 4350 -

FORECAST 4283

1/24/2014

12

Weighted moving average


MONTHS JANUARY FEBRUARY WEIGHTS 2 3 SALES 4200 4300 WEIGHTED SALES 8400 12,900

MARCH
TOTAL

5
10

4350

21,750
43050

Weighted forecast for April = 4305


1/24/2014 13

Major factors that influence demand forecasting Product promotions -one product stealing demand of another product (tooth powder and tooth paste, motor car and motorbikes) Lead times Strong forecasting methods for long lead-times
1/24/2014

Simpler methods for shorter lead-times

14

Influence of product variants on each other


Full shirts and half shirts, shirts and T-shirts,

different models of same product

1/24/2014

15

Appropriate technique for forecast

Take the dimensions of forecast into account to


determine forecasting method.

Geographical area
Product groups (consumer/industrial)

Customer groups (income, age etc)

1/24/2014

16

Take criteria into account


accuracy

time horizon
data availability experience of the forecaster

1/24/2014

17

Performance measures and error measures to


use forecast accurately:

lead time is a performance measure


Forecast accuracy is required to be highest at the

end of this lead-time.


Difference between forecast and actual in the past should be measured for estimating error.
1/24/2014 18

Forecast Approaches Top-Down Approach (decomposition approach)

A national level forecast for SKU of company performance pattern of locations in the past forecast for various locations demand is assumed to be uniform across the
national market
1/24/2014 19

Bottom-Up Approach (decentralized approach)

Forecast for individual locations

Cumulative forecast for company at national


level

1/24/2014

20

You might also like