COMM 652 Business Analytics

Learning Objectives
Predictive Analytics At the end of today you will be able to:
 Explain why managers forecast
for Management
 Briefly describe forecasting steps

 Differentiate between Explanatory and

Extrapolation methods
 Differentiate between the different
components of a time series
– Forecasting Game!
 Be able to calculate 2 methods for forecasting
a stationary series (manually and Excel)
 Next Class: start into Trends and Seasonality

1 2

Team Exercise (5 minutes) What is a Forecast?

Without the use of technology (laptops,
phones, etc):  A prediction of the future
 Define “Forecasting” “Predictive Analytics”  fore = before + cast = throw
 Come up with 3 examples of forecasting  Literally planning before you throw.
(preferably in business)
 Why forecast? Who forecasts?
 What do companies forecast?
Forecasting
 What do they need to forecast?
 “same as last week” is a simple forecast
 What do companies do with such forecasts?
 Forecasting can be extremely complex
 How do they forecast?

3 4

What type of organizations

Planning vs. Forecasting Forecast? Who?
 Forecasting
– A process of predicting what the future will be like
– Forecasts are used as inputs to the planning
process

 Planning
– A process of determining how to deal with events
in the future

5 6

© Johnson, M. 2024. All Rights Reserved. 1

COMM 652 Business Analytics

What Can Be Forecasted? What Can Be Forecasted?

The External Environment The Internal Environment
 National Economy  Marketing

 Customer behaviour  Production

 Other producers  Financial Position

 Company distributors Executive goal-setting and planning is

intricately linked to forecasting

7 8

What Can Be Forecasted?

Why Forecast?
 Demand (Sales/Activities)
– By industry, by market share, by product,  It’s fun
competition, price  To look smart
 Costs  Most importantly: To make better decisions
– Raw materials, wages, semi-finished goods,
interest rates
 What are the effect of bad forecasts?
 Technology
– Excess costs – too much staff or stock
 Social, Political and Economic Trends
– Poor service – waiting lines or stock outs
 Projects

9 10

Why Forecast? Why Forecast?

 Scheduling  Forecasts are necessary for effective
 Acquiring resources – too many vs. too few decision making
 Determining resource requirements –Forecasting, planning and control are interrelated

 Managers should be proactive

– Anticipate problems and prepare

11 12

© Johnson, M. 2024. All Rights Reserved. 2

COMM 652 Business Analytics

General Truths
Forecasting Realities
Any forecast you make is almost certain to  Bad News: Forecasts are always wrong
be wrong.  Good News: Forecasts are like grenades
Then why bother?
– You only have to be close!

 Use forecasting methods if you can make better

decisions with it, than without it.
 Some elements you can influence, some you can’t
 Scientific methods improve forecasting
 Constantly monitor the quality of your forecasts

13 14

How are Forecasts Used?

General Truths
Manufacturing Planning
 Some things are easy, some are hard
 Broader forecasts are more accurate
– E.g. Forecast sales of shirts vs. sales of
Oxford-style, cotton, size 17/40, brown shirts.
 Shorter is easier than longer
– E.g. Forecast sales next month vs. next year

15 16

Types of Forecasting Methods 2. Time Series Forecasting

 Also known as “Extrapolation” models
 Chapter 11 Forecasting Handout in Class 3
Categories: folder of Moodle:
1. Qualitative – judgemental forecasts  Moving Average
 Weighted Moving Average
2. Time Series Methods – Intrinsic Methods
 Exponential Smoothing
3. Causal Methods – Extrinsic Methods  Trend Models
 Seasonal Models
 Trend and Seasonal Models

17 18

© Johnson, M. 2024. All Rights Reserved. 3

COMM 652 Business Analytics

2. Time Series Forecasting 3. Causal Methods

 Time is the independent variable  Also known as Explanatory or Econometric
 Forecasting is based on past data – Based on extrinsic factors that might influence
the quantity being forecasted
 One basic assumption:
– Regression based
– Thefuture will behave approximately like
the PAST
 To forecast future sales we use?
 Problem?
 Can be “wiped out” by unexpected events
e.g., Covid-19 panademic

19 20

© Johnson, M. 2024. All Rights Reserved. 4