FORECASTING

Nikhil Ghag
Business Analyst
TheGreenBillions Limited
WHAT IS FORECASTING
 Forecasting is by far the most important and frequently used application of predictive analytics
since it has significant impact on both top line and bottom line of an organization.
 Every organization prepares long-range and short-range planning for the organization and forecasting
demand for product and service is an important input for both long-range and short-range planning
 Different capacity planning problems such as manpower planning, machine capacity,
warehouse capacity, materials requirements planning (MRP) will depend on the forecasted
demand for the product/ service.
 Budget allocation for marketing promotions and advertisement are usually made based on
forecasted demand for the product
INTRODUCTION TO FORECASTING

 Boeing 747-400 has more than 6 million parts and several thousand unique parts (Hill, 2011).
Forecasting demand for spare parts is important since non-availability of mission critical parts
can result in aircraft on ground (AOG) which can be very expensive for airlines.
 Amazon.com sells more than 350 million products through its E-commerce portal. Amazon
itself sells about 13 million SKUs and has more (about 2 million) retailers selling their
products throughAmazon
 Demand for products and service is not the only application of forecasting, even manpower
planning requires the use of sophisticated models.
 Many products may have intermittent demands, that is, gap between two demands can be long
and the gap itself may be random
ROLE OF FORECASTING IN A
SUPPLY CHAIN
 Used in Push and Pull
 Production: Scheduling, inventory, aggregate planning
 Marketing: Sales force allocation, promotion, new production
introduction.
 Finance: Plant/equipment investment, budgetary planning.
 Personnel : workforce planning, hiring, layoffs.
 All of the these decisions are interrelated.
CHARACTERISTIC OF
FORECAST
 Forecast are always wrong. Should include expected values and measure of error.
 Long term forecast are less accurate that short term forecast(forecast horizon is important)
 Aggregate forecast are most accurate than disaggregate forecast.
FORECASTING METHODS
 Qualitative: primary subjective, rely on judgement and opinion
 Time series: use historical demand only
 Static
 Adaptive
 Casual: Use the relationship between demand and some other facto to develop forecast.
 Simulation:
 Imitate consumer choice that give rise to demand.
TIME-SERIES DATA AND COMPONENTS OF TIME-SERIES
DATA

 Trend Component (Tt): Trend is the consistent long-term upward or downward movement
 Seasonal Component (St): Seasonal component (measured using seasonality index) is the
repetitive upward or downward movement (or fluctuations) from the trend that occurs within a
calendar year such as seasons, quarters, months, days of the week.
 Cyclical Component (Ct): Cyclical component is fluctuation around the trend line that
happens due to macro-economic changes such as recession, unemployment, etc.
 Irregular Component (It): Irregular component is the white noise or random uncorrelated
changes that follow a normal distribution with mean value of 0 and constant variance.
TREND IN TIME-SERIES DATA.

 The time-series data can be modelled as an addition of the above components or product of the
above components. The additive time-series model is given by
QUANTITATIVE METHOD
Simple Average- Random Moving average method –
values when there is trend- five
month moving average, three
month moving average.
Weighted moving average Exponential Smoothing Equation of Exponential
method method – include all the data – smoothing method.
weighted recent observation
much more heavily than very
old.
PROBLEM
SOLUTION
PROBLEM
AUTO REGRESSION, ARIMA
AND ARMA
 What is auto regression?
 How to identify the order?
 What are ACF and PACF?
 What is stationarity and seasonality?
 Equation of Auto regression.
 Pearsons's correlation coefficient [-1,1]
STATIONARITY AND
SEASONALITY
ARMA AND ARIMA
FORECASTING TECHNIQUES AND FORECASTING
ACCURACY

 There are many forecasting techniques developed based on different logics. Simple techniques such as
moving average and exponential smoothing predict the future value of a time-series data as a function of
the past observations.
 Usually, many different forecasting techniques such as moving average, exponential smoothing, and
ARIMA are used for forecasting before selecting the best model. The model selection may depend on the
chosen forecasting accuracy measure. The following four forecasting accuracy measures are frequently
used:
1. Mean absolute error
2. Mean absolute percentage error
3. Mean squared error
4. Root mean square error
MEAN ABSOLUTE ERROR (MAE)

 Mean absolute error (MAE) is the average absolute error and should be calculated on the validation data
set. Assume that the validation data has n observations and forecasting is carried out on these n
observations using the model developed. The mean absolute error is given by

 It is one of the popular forecasting accuracy measures used by practitioners since it expresses the average
error in percentage terms and is easy to interpret.
MEAN ABSOLUTE PERCENTAGE ERROR (MAPE)

 Mean absolute percentage error (MAPE) is the average of absolute percentage error. Assume that the
validation data has n observations, and the forecasting is carried out on these n observations. The mean
absolute percentage error is given by

It is one of the popular forecasting accuracy measures used by practitioners since it

 It is
expresses the average error in percentage terms and is easy to interpret.
MEAN SQUARE ERROR (MSE)

 Root mean square error (RMSE) is the square root of mean square error and is given by

 Lower MSE implies better prediction. However, it depends on the range of the time-series data.
ROOT MEAN SQUARE ERROR (RMSE)

 Root mean square error (RMSE) is the square root of mean square error and is given by

 RMSE along with MAPE are two most popular accuracy measures of forecasting. RMSE is the standard
deviation of errors or residuals. In 2006, Netflix, the movie portal, announced a competition with a prize
money worth one million dollars to predict the rating on a 5-point scale likely to be given a customer for
a movie2 (source: Wikipedia). The participants were given a target RMSE of 0.8572 to qualify for the
prize.