Sie sind auf Seite 1von 31

DEMAND FORECASTING

TECHNIQUES
Importance/ uses
 Demand forecasting is the process of estimating future demand for
product, industries, or industry sectors based on demand data of
previous years or using opinions of concerned experts in the
industry.
 Uses of Demand forecasting
 It helps to predict future demand. So company can plan
 future production capacities,
 purchase of raw materials
 marketing plans,
 HR recruitment,
 Financing methods etc.
Features of a good forecast

 Must give accurate results


 Must be simple to calculate and understand
 Must be low cost method
 Less time consuming, fast.
 Flexible to use in different time periods, different industries.
 LIMITATIONS
 Lack of availability of past data.
 Unpredictable change in external factors, peoples taste etc.
 Non- availability of experts who can give their opinion.
 Lot of time and money required for survey method etc.
2 types of forecasting methods
Qualitative methods, Quantitative methods
 Qualitative methods are based on Opinions of people like Customers,
Salesmen, Experts in the field etc.
 Quantitative methods use quantitative data and statistical methods.
 Qualitative Quantitative
 Consumer opinion ( survey) Time series
 Sales force composite opinion Moving average method
 Expert opinion/ jury of Executive Trend projection/ Least square
 Delphi method Barometric method ( Index)
 Exponential smoothing
 Trend projection can be done using Graphical method or Least square
regression method. Exponential smoothing gives more weightage to recent
data.
Examples
 Consumer opinion- Horlicks can find the future demand by conducting
a survey among households. Apollo tyres can find future demand by
asking their customers Maruti, Hyundai, Tata motors etc.
 Salesforce composite- a company can hold a meeting of all sales team
and ask them to give future demand opinion. Take total.
 Expert jury opinion- a telecom company can ask a team of experts
about demand for 5G, 6G or demand for Electric car etc.
 Delphi- same as Expert jury, but a coordinator collects the opinions
using several rounds.
 Time series, Moving averages, trend projection, exponential
smoothing- if a company has sales data for last 10 years, it can forecast
demand for next year using these methods.
 Barometric method- a furniture company can fore cast demand based
on an index of House starts or marriages taking place in a year.
Delphi method
 It is a group opinion collection method.
 Members of the group are experts in the field ( 10/15)
 There will be a coordinator or super expert in the group to coordinate
the opinions
 In first round, each expert member gives his opinion in a secret slip of
paper.
 Then the Coordinator circulate the slips among all members. So each
member comes to know what opinion others have given. Based on this
knowledge, they may modify their opinion in the next rounds.
 There will be several rounds of opinion collection and circulation.
 Finally all members will converge on a single opinion. Then the exercise
is stopped.
TIME SERIES METHOD
 THIS METHOD FORECASTS FUTURE DEMAND BASED ON PAST DATA OF
SEVERAL YEARS OR PAST TREND
 When using past data, we must be aware that 4 types of fluctuations or
variations in data are possible. So the influence of these fluctuations are
to be removed to get a correct projection.
 Long term secular Trend ( T)
 Cyclical movements ( C) – within trend there are cycles
 Seasonal variations ( S) – variations within an year
 Random movements( R) – due to unpredictable events like war, drought
etc
 So Total projected demand is to be corrected for each of these factors.
 Y = T + C + S + R or T x C x S x R
 Additive form or Multiplicative form can be used.
Business Cycle
Peak
Peak

Trough
Expansion
 During a period of expansion:
 Wages increase
 Low unemployment
 People are optimistic and spending money
 High demand for goods
 Businesses start
 Easy to get a bank loan
 Businesses make profits and stock prices increase
Contraction
 During a period of contraction:
 Businesses cut back production and layoff people
 Unemployment increases
 Number of jobs decline
 People are pessimistic (negative) and stop spending
money
 Banks stop lending money
Moving Average Method
 Average demand of past 3 months or 5 months is used as the Demand for
next month. ( 3 months moving average or 5 month moving average)
 Eg.
 Month Jan Feb Mar Apr May Jun July
 Demand 100 120 110 130 150 120. ?
 Find the Demand for July using 3 months moving average
 Ans. Demand for July will be average of Apr, May, June.
(130+150+120)/3 =
 Q. Find demand for July using 5 months moving average.
Weighted moving average
 Weights are assigned to recent data. Multiply the data by weightage and take
total.

 Month Jan Feb Mar Apr May Jun July


 Demand 100 120 110 130 150 120. ?
 Find the Demand for July using 3 months weighted moving average.
Weightages are given as 0.5, 0.3, 0.2
 Ans
 Demand for July = 0.5 x 120 + 0.3 x 150 + 0.3 x 130 = 144.
What if we use a 3-month simple moving average?

AJun + AMay + AApr


FJul = = 1,227
3

What if we use a 5-month simple moving average?

AJun + AMay + AApr + AMar + AFeb


FJul = = 1,268
5
Exponential smoothing
LEAST SQUARE METHOD/ REGRESSION
EQUATION/CASUAL METHOD
 This method uses mathematical formula to predict future demand from
past data. It tries to fit a Trend line or Regression line on the data
available.
 Equation for a Trend line or Regression line is, Y = a + b X where a and
b are constants. Y is the Dependent variable ( Eg. Sales) and X is the
Independent variable ( Eg. Advertisement).
 If data of last 10 year Sales and Advertisement is given, we can forecast
the Sales for 11th year or any future year, if Advertisement amount is
known.
 Graphically, a is the Y-intercept of the Trend line and b is the Slope of
the trend line.
Trend line / Regression line Y = a + b X
Formula to find b and a values
 Data given- 10 years Sales data Y and 10 years Advertisement data
X.
 Find Average of Y ( Y bar) and Average of X ( X bar),
 Find Sum ( Y- Y bar) ( X – X bar), Sum (X- X bar)2
 Sum Product variance / Sum square of variance.
 b = Sum ( X- X bar) ( Y-Y bar) divided by Sum (X- X bar)2
 Then find value of a by using equation a = Y bar – b X bar
 Thus if we get values of a and b, we can write Trend line/
Regression line as
 Y = a + b X or Sales = a + b x Advertisement. If a and b are known,
we can Forecast sales for any future year , given Advertisement
value.
Formula for a and b.

b = product of variance / square of variance.


University- Numeric problem.
Develop a Trend equation and predict the sales
for 2017 and 2018

 Sales for a company from 2012 to 2016 is given as


 Year sales
 2012 85
 2013 92
 2014 89
 2015 105
 2016 108

 Forecast the Sales for 2017 and 2018.


 Clue- year may be taken a 1,2,3,4,5
Equations to be solved to find a and b
 Y=a+bX
 Sales = a + b x Year

 ∑Y=na+b∑X
 ∑ XY = a ∑ X + b ∑ X2
 Solve these 2 equations using the given data and get the value of a and b
 Substitute in Y = a + b X
 Substitute X as 6 to get Sales for 2017 and substitute X as 7 to get Sales of
2018.
Calculations
Year X Sales Y Year X X2 XY
2012 85 1 1 85
2013 92 2 4 184
2014 89 3 9 267
2015 105 4 16 420
2016 108 5 25 540
Total 479 15 55 1496

Apply formula
479 = 5a + 15 b
1496 = 15 a + 55 b
Solve to find values of a and b.
Substitute a and b in Trend equation Y = a + b X
use X value 6 to find sales of 2017, use X value 7 to find
sales of 2018.
Univ question
 Develop a Trend line/ Regression line and find the sales for the 6th
period.
 Period Sales
 1 120
 2 140
 3 120
 4 150
 5 180

 Important note- even if period is given in years, 2012, 2013, 2014,


2015, 2016, for calculation we can take years/periods as 1,2,3,4,5.
Solution
x y X-X bar Y – Y bar (x-x bar) ( y – y bar) (x-x bar)2
1 120 -2 -22 44 4
2 140 -1 -2 2 1
3 120 0 -22 0 0
4 150 1 8 8 1
5 180 2 38 76 4

15 710 130 10
15/5= 710/5 b= 130/10 = 13
3 = 142

b= 13, now substitute value of b in a = Ybar- bXbar to get a = 103


a = 142 – (13 x 3) = 103
Trend equation is. Y = 103 + 13 X
substitute X = 6 to get sales of 6th period. Ans. 181
Formula for a and b.

b = product of variance / square of variance.

Das könnte Ihnen auch gefallen