Forecasting in operations management: meaning, time horizons, demand patterns, qualitative and quantitative methods, and a worked moving average and exponential smoothing example with MAD.
Updated 10 min read
Forecasting is the process of estimating future demand, sales, needs or trends from past data and present conditions, so that a business can plan ahead. A college predicting next year's admissions and a bakery estimating how many cakes it will sell this weekend are both forecasting.
In operations management almost every plan starts with a forecast. It tells a business how much to produce, how many workers are needed, how much raw material to buy and how much stock to keep. A poor forecast leads to overproduction and waste, or to stock-outs and lost customers; a good one lowers cost and uncertainty across the whole organisation.
Forecasting is the process of estimating future events, especially demand, for planning and decision-making purposes. A forecast is a statement about the future; it is not a plan or a target. The plan (how much to make) is decided after the forecast (how much customers are likely to buy).
Every forecast has three parts: what is being forecast (units, rupees, hours), for which period (the forecast horizon), and how accurate it is likely to be.
Covers days, weeks or a few months (usually up to about three months). It is used for daily production, scheduling workers, ordering raw materials and setting job assignments. Example: a restaurant predicts weekend customer demand.
Covers roughly three months to two years. It is used for sales planning, production planning and budgeting, and aggregate planning of workforce and inventory. Example: a garment maker estimates demand for the coming festival and winter seasons.
Covers several years. It is used for expansion plans, new product decisions, investment planning, capacity increases and facility location. Example: a school estimates student strength for the next three years before building new classrooms.
Horizon
Typical period
Used for
Usual methods
Short term
Up to about 3 months
Scheduling, purchasing, staffing
Moving average, exponential smoothing
Medium term
3 months to 2 years
Budgeting, aggregate planning
Trend projection, regression
Long term
2 years and more
Capacity, location, new products
Qualitative methods, regression
Short-term forecasts are generally more accurate than long-term ones, because fewer things can change in a short period.
These rely on judgement, opinion, experience and market knowledge. They are used when past data is limited, for example for a new product, or when conditions are changing fast.
Jury of executive opinion: senior managers pool their views.
Sales force composite: each salesperson estimates sales in their territory, and the estimates are added up.
Market survey (consumer survey): customers are asked about their buying intentions.
Delphi method: a panel of experts answers questionnaires in several anonymous rounds, with feedback between rounds, until their views converge.
Expert opinion: specialists outside the firm are consulted.
Example: a company asks experts whether a new course will have enough demand.
Weighted moving average, with weights wi that add up to 1:
Ft+1=∑wiAi
Simple exponential smoothing, with smoothing constant α between 0 and 1:
Ft+1=Ft+α(At−Ft)
Mean absolute deviation (MAD), a common accuracy measure:
MAD=n∑∣At−Ft∣
Here At is actual demand and Ft is the forecast for period t. A larger α makes the forecast respond faster to recent changes; a smaller α gives a smoother forecast.
Worked example: moving average and exponential smoothing#
Suppose a bakery recorded the following monthly cake sales and wants a forecast for Month 9.
If the bakery gives weights of 0.5, 0.3 and 0.2 to the latest, second-latest and third-latest months:
F9=0.5(165)+0.3(160)+0.2(145)=82.5+48+29=159.5
The weighted forecast (about 160 cakes) is higher because it leans more on the rising recent months.
Step 3: exponential smoothing with alpha = 0.3#
Assume the forecast for Month 1 was 120. Then:
F2=120+0.3(120−120)=120.00
F3=120+0.3(130−120)=123.00
F4=123+0.3(125−123)=123.60
F5=123.60+0.3(140−123.60)=128.52
F6=128.52+0.3(150−128.52)=134.96
F7=134.96+0.3(145−134.96)=137.97
F8=137.97+0.3(160−137.97)=144.58
F9=144.58+0.3(165−144.58)=150.71
The exponential smoothing forecast for Month 9 is about 151 cakes.
Step 4: compare accuracy using MAD (Months 4 to 8)#
Month
Actual
3-month MA
Absolute error (MA)
Exp. smoothing
Absolute error (ES)
4
140
125.00
15.00
123.60
16.40
5
150
131.67
18.33
128.52
21.48
6
145
138.33
6.67
134.96
10.04
7
160
145.00
15.00
137.97
22.03
8
165
151.67
13.33
144.58
20.42
Total
68.33
90.37
MADMA=568.33=13.67MADES=590.37=18.07
Both forecasts trail the rising demand; the 3-month moving average trails it less in this example.
Interpretation: the moving average has the lower MAD, so it was more accurate for this data. Both methods lag behind a rising trend, and exponential smoothing with a small α of 0.3 lags the most. For data with a clear trend, a larger α or a trend-adjusted method would usually do better.
In summer, demand for ice cream is high; in the rainy season it is lower. The shop forecasts demand for each season and uses that forecast to decide how much stock to keep, how many workers are needed and how much milk and sugar to buy. Recognising the seasonal pattern is what makes the forecast useful.
5-mark answer. Forecasting is the process of estimating future demand or future events for planning purposes. It is important in operations management because it helps in production planning, inventory control, manpower planning and budgeting. Forecasts may be short-term, medium-term or long-term. The main methods are qualitative (such as the Delphi method and market surveys) and quantitative (such as moving averages and exponential smoothing). Good forecasting reduces uncertainty and improves decision-making.
2-mark answer. Forecasting is the process of predicting future demand or future events for planning and decision-making.
Memory line. Forecasting means looking ahead so that planning is better.
What is the difference between qualitative and quantitative forecasting?#
Qualitative methods rely on judgement and opinion and are used when data is scarce, such as for a new product. Quantitative methods use past numerical data and mathematical models and are used when reliable history exists.
Which value of alpha should be used in exponential smoothing?#
A higher α (for example 0.5 or more) reacts quickly to recent changes but also to random noise. A lower α (0.1 to 0.3) gives a smoother, more stable forecast. The best value is usually the one that gives the lowest error, such as MAD, on past data.
A moving average gives equal weight to older periods, so when demand is rising steadily the average is always pulled down by the older, lower values. Weighted averages and trend methods reduce this lag.
What is the difference between a forecast and a plan?#
A forecast estimates what customers are likely to demand. A plan decides what the firm will actually do, such as how much to produce, taking the forecast, capacity and costs into account.
By comparing forecasts with actual results. Common measures are mean absolute deviation (MAD), mean squared error (MSE) and mean absolute percentage error (MAPE). A lower value means a more accurate method.
