Introduction to Forecasting

1
 In an manufacturing organization
Product is ready for full scale manufacturing.

What must be the next step?

2
 Forecasting- The Primary function of
operations
How much to Produce?

What is the number of product that can be

manufactured & sold in the market?

Thus we have to Forecast.

3
 Importance of Forecast to an
organization
Produce without forecast:
1- No market for product
2-Big market for Product

1-It can influence the success or failure of any

organization.

2-It helps in planning activity and decision making

activity of any organization.

Many decision based on forecasting

4
 Importance of Forecasting to supply
chain
It is very important as rest of the entire planning of your
supply chain depends on Forecasting.

Production- Aggregate Planning, inventory

Marketing- Sales force allocation, Promotions,
Finance – Plant investment, equipment
investment ,Budgeting

Forecasting actually provide input for all these functions

/activity.

5
 Ex: Four wheeler Manufacturing, Hospital, Academic Institute,
Ac/Gyser etc.

Forecasting of 10000 Four wheeler in next six month

Tires
Steering
Seat covers
Gearbox etc.

Work force can produce only 8000 Four wheeler

Material planning
Financial planning
Resource planning

6
Forecast
Art
Science
Data Analysis

7
Accurate Forecast
Matching actual demand
Precise and Accurate
Very difficult
Mathematical tools
Previous data available

8
 Forecast is not always accurate
Wrong Weather forecast

So some times forecast are not accurate and thus companies need to be dynamic in nature to adjust this
type of fluctuation in demand .

9
Thus forecast may be fairly accurate but we can not say that it will be absolutely accurate.

Persons who can make accurate forecast are precious.

If forecast = X
Actual demand = X ± Delta * X
Minimize Delta * X

10
 Forecasting Definition
Forecast is the process of estimating the future demand in terms of quantity ,timing and location
of desired products and services .

Quantity
Timing
Location

Thus Forecasting is the process of estimating the future demand. Based on past demand
information.

11
Forecast is a Magic Number
Material planning
Resource Planning
Production Planning
Aggregate Planning

Planning activity can be done well informed manner. Thus helps in overall planning and decision
making.

12
 Importance of Forecasting is even more
relevant now a days
Due to reduce Product life cycle

Business environment is quite volatile

Technology is changing every now and then

Thus importance of forecasting is very high

13
Short term 0-3 months
Medium term 6 months to 18 months
Long term forecast more than that

14
 A forecast is as good as the information included in the past
data
Relying on the past data

Previous data is very important.

So past data should not be confusing it should strongly related to what you expect to see in
future.

15
We need to check
What is past data
Whether it is relevant in today’s context
Can we use it to make forecasting

16
Can we predict the new e bike demand based on data on ICE bike?

So be very careful in the selection of data.

Because Market segment may be entirely
1-HOW MUCH THE TWO MARKET HAVE IN COMMON?
2-IS THE CUSTOMER BASE SAME?
different Terms and condition may be entirely
ONE CAN NOT USE ONE PARTICULAR PRODUCT DATA TO OTHER SPECIFIC PRODUCT. different

Month Bike e-Bike Age wise segment

Jan 20k
Feb 15k Type of customer focusing on bike may be
different
Mar 18k
April 14k
May 17k

17
Characteristics of Forecast
They are usually wrong.
Aggregate forecasts are more accurate.
The longer the forecast horizon, the less accurate the forecast will be.

18
19
What should be consider when looking at past demand
data
Trends
Seasonality
Cyclical elements
Random variations

20
21
 Types of Demand Forecasting
Methods of demand forecasting
Demand Forecasting

Qualitative Analysis Quantitative Analysis

Customer Survey Sales Force Composite

Time Series Analysis Causal Analysis

Executive Delphi
Opinion Method Simple Moving Simple Trend Analysis
Average Exponential
Smoothing

Past Analogy
Holt’s Double Winters’s Triple
Exponential Exponential
Smoothing Smoothing

Forecast by Linear
Regression Analysis

22
Quantitative method
Quantitative forecasting methods (time series methods and regression). Using
objective(quantitative) forecasting methods, one makes forecasts based on past history.

Time series forecasting uses only the past history of the series to be forecasted, while
regression models often incorporate the past history of other series.

23
Quantitative method
1- Time series Model
Naïve approach
Moving average method
Exponential smoothing Method

2-Causal models
Trend projection
Linear regression Analysis

24
Naive approach
Quick and easy to use
Has no cost and easy to understand

The simplest way of forecast is to assume that forecast of demand in next period is equal to
the actual demand in the most recent period(i.e. current period).

Ex: Using yesterday’s sales as tomorrow’s sales forecast

They are so simple that they result in substantial forecast error

25
26
Techniques for averaging
Historical data contain a certain amount of random variation or noise. It is desirable to
completely remove any randomness from data and leave only real variations such as change in
demand.

Averaging techniques smooth fluctuations in a time series so that the forecast can be based on
an average , to exhibit less variability than the original data.

Moving Average
Exponential smoothing

27
Moving Averages
A moving average is used to smooth the trend in a time series.
It is the basic method used in measuring seasonal fluctuation.
To apply a moving average, the data needs to follow a fairly linear trend and
have a rhythmic pattern of fluctuations.
This is accomplished by “moving” the mean values through the time series.

28
Moving Average Example
The graph of sales fluctuates
but the moving average
removes the cyclical and
irregular fluctuations leaving
an upsloping trend line.

29
3- and 5-Year Moving Average Example
This is a table and graph of production numbers along with 3-year and 5-year
moving totals and moving averages. Note, moving averages do not always
result in a precise line.

30
A Weighted Moving Average Example
Cedar Fair operates eleven amusement parks, three outdoor water parks, one indoor water park, and
five hotels. Its combined attendance (in thousands) for the last 20 years is given in the following
table. A partner asks you to study the trend in attendance. Compute a three-year moving average
and a three-year weighted moving average with .2, .3, and .5 weights.

Here is the 3-
year moving
average.

31
A Weighted Moving Average Example
Continued
Notice the weighted moving average follows the data more closely than the
moving average.

32
33
34
35
36
37
38
 Exponential function

Exponential functions have the form f(x) = bx, where b > 0 and b ≠ 1. ... Where b is a constant
and x is a variable. An example of an exponential function is the growth of bacteria. Some
bacteria double every hour. If you start with 1 bacterium and it doubles every hour, you will
have 2x bacteria after x hours. This can be written as f(x) = 2x.

Exponential Growth Exponential Decay

In Exponential Growth, the quantity In Exponential Decay, the quantity
increases very slowly at first, and then decreases very rapidly at first, and then
rapidly. slowly
39
Difference between exponential and
moving average
Consider a system in which the demand for 300,000 inventory items is forecasted each month
using a 12-month moving average. The forecasting module alone requires saving 300,000 * 12 =
3,600,000 pieces of information.

If exponential smoothing were used, only 300,000 pieces of information need to be saved.

40
Exponential smoothing

Simple moving average and weighted moving average require a lot of data.

Exponential smoothing is based on the premise that the latest occurrence of data is more
indicative of the future than the past occurrence. this method gives exponentially decreasing
importance to older data and require very little data.

It is based the actual data of the previous period and smoothing constant alpha which lies
between zero and one. Exponential smoothing is the most commonly used forecasting
technique it is surprisingly accurate and easy to use.

41
 Formula for Simple Exponential
smoothing

Forecast for period t = Forecast for period (t-1)+ alpha (Forecast error in period (t-1)

(At-1 - Ft-1 ) is the forecast error for period t-1

ALPHA is the smoothing parameter that defines the weighting and should be greater than 0 and less than 1.

42
The equation can also be written

The damping factor is a corrective factor that minimizes the instability of data collected
across a population. The default damping factor is 0.3.

43
44
45
Linear Trend
 The long-term trend of many business series, such as sales and production,
often approximates a straight line

46
Least Squares Method Example
The sales of Jensen Foods, a small grocery chain located in
southwest Texas, for 2012 through 2016 are in the table below.
Determine the regression equation. How much are sales increasing
each year? What is the sales forecast for 2018?

47
Methods for forecasting stationary time series. We consider two forecasting methods when the
underlying pattern of the series is stationary over time: moving averages and exponential
smoothing.

Methods for forecasting series with trend(Non stationary). When there is an upward or downward linear
trend in the data, two common forecasting methods are linear regression and double exponential
smoothing via Holt’s method.

48
Causal Models
Cause -Effect
Causal models such as linear regression ,
incorporate the variables or factors that
might influence the quantity being forecast.
In linear regression the dependent variable is
related to one or more independent variable
by a linear equation.

Demand of sales forecast is dependent

variable and other factors that effect demand
are independent variables (causal variables).

49
Linear Regression

A model which uses what is called the least square method to identify the relationship between
dependent variable and one or more independent variable that are present in his set of
historical observations.

In simple regression there is only one independent variable in multiple regression there is more
than one independent variable if the historical data set is time series the independent variable
is the time period.

50
Linear Regression

51
52
53
CPFR

54
55
56
◦
Thankyou

57