Forecasting quiz OSCM

What is a forecast?

A prediction of future events used for planning purposes

what are the five basic patterns of most demand time series

1. horizontal

2. trend

3. cyclical

4. seasonal

5. random

horizontal time series

constant with no longterm upward or downward movement

trend

consistent upward or downward line over time

seasonal

regular, changes with seasons

cyclical

wavelike movements lasting more than a year usually linked to economic cycles

random

unpredictable variations caused by unexpected events

Key Decisions in Making a Forecast

1. deciding what to forecast

2. level of aggregation

3. units of measurement

deciding what to forecast

define the target variable and its scope

level of aggregation

determines the degree of detail in the forecast. aggregate (as a whole) or disaggregate (as a unit)

units of measurement

specifies how demand is expressed

type of forecasting techniques

1. judgment methods

2. causal methods

3. time series analysis

4. trend projection with regression

judgement methods

relies on expert opinions

causal methods

predicts future demand based on cause and effect. y=mx+b

time series analysis

use historical demand data over time to predict future values

trend projection with regression

applies regression analysis to historical demand data over time

Judgement methods 1-4

• used when there is little to no historical data or if the product is new

1. Executive Opinion

2. Sales Force Composite 

3. Delphi Method

4. Consumer Market Surveys

Executive Opinion

High-level managers meet to pool knowledge and establish a forecast

• longe-range planning

Sales- Force Composite 

Employees close to the customer provide estimates

• used for short term and specific forecasts

Consumer Market Surveys

getting opinions from potential customers

• best for new product introductions

Delphi Method

getting people from every department to come up with a forecast, then coming up with a consensus

• best for long-term strategic forecasts to mitigate groupthink

forecast error

actual demand- forecast demand

Common Forecast Errors

• CFE

• AFE

• MAD

• MSE

• MAPE

• standard devation

Causal forecasting using Linear Regression

• 2 variables, one dependent and one independent

• making a graph and comparing the correlation

time-series method of forecasting

uses historical data to try and predict future data

1. naive 

2. moving average

3. weighted moving average

4. exponential smoothing

5. trend projection

correlation coefficient

how strongly two variables move together/ are related to each other

• varies -1 to 1

coefficient of determination

how much of the change in one variable is explained by the other

• %

• correlation squared

2 types of forecasting techniques

1. qualitative (judgment)

2. quantitative (data)