ADVANCED BUSINESS FORECASTING - NKD Group

1 ADVANCED BUSINESS FORECASTING Definition FORECASTING involves making the best possible judgment about some future event. In other words, forecasts are numerical estimates of an event for some future date that can be achieved with a specified level of support and are reproducible. "I often say that when you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a very meagre and unsatisfactory kind." William Thomson, Lord Kelvin, 1824-1907 "If we could first know where we are, then whither we are tending, we could then decide what to do and how to do it.

2 " Abraham Lincoln, 1809-1865 The elements that come into play in all FORECASTING methods is the concept of the future and time; uncertainty; and reliance on historical data. MELec6_6: FORECASTING Page: 2 MBA555 Class Notes Prepared by: These notes contain copyrighted information Dr. Nina Kajiji Do not quote or copy without permission Last Update: Nov 9, 2009 Major Types of FORECASTING Methods Subjective Methods Sales Force Composites Customer Surveys Jury of Executive Opinions Delphi Method Quantitative Methods Exponential smoothing family ARIMA Artificial Neural Networks (ANN) Elements of FORECASTING Source Of Data Time Domain MELec6_6: FORECASTING Page: 3 MBA555 Class Notes Prepared by: These notes contain copyrighted information Dr.

3 Nina Kajiji Do not quote or copy without permission Last Update: Nov 9, 2009 Components of a Time Series Secular Trend The growth of the economic system is tied to the growth of BUSINESS and industry. Based primarily in growth of the population. Trend can be linear ttYa bx= + + or nonlinear as shown in the graph. In a nonlinear trend: o Data that increases by a constant amount each successive time period o Data that increases by increasing amounts at each successive time period. o Data that increases by an equal percentage at each successive time period (easy to linearize by the use of logarithms). MELec6_6: FORECASTING Page: 4 MBA555 Class Notes Prepared by: These notes contain copyrighted information Dr.

4 Nina Kajiji Do not quote or copy without permission Last Update: Nov 9, 2009 Seasonal Variation Variation in BUSINESS and economic activity that results from changing seasons. Periodic fluctuations in consumer spending ===> Periodic Sales ===> Periodic production Cyclical Fluctuations Cyclical fluctuations are not very predictable. BUSINESS cycles -- Expansion and Contraction Erratic Fluctuations Are the fluctuations really erratic (chaotic)? Here is where our work with high-frequency data (daily level or greater frequency) and new modeling / FORECASTING methods will prove to be most useful. That is, time-series that appeared to have no economic value are now being modeled quite successfully with new methods like ANNs.

5 MELec6_6: FORECASTING Page: 5 MBA555 Class Notes Prepared by: These notes contain copyrighted information Dr. Nina Kajiji Do not quote or copy without permission Last Update: Nov 9, 2009 Quantitative FORECASTING Methods Charting Approaches (read) Moving average Exponential Smoothing Artificial Neural Networks Charting Approaches Arithmetic Charts (Scatter Plots) Definition The purpose of the arithmetic charts is to show actual movement of the time series from one period to the next. Unless the axis markings (scaling) are the same it is not possible to compare and interpret two different charts. A chart drawn on the arithmetic scale compares the amounts of change.

6 Data with a wide dispersion in values may not be accurately reflected. MELec6_6: FORECASTING Page: 6 MBA555 Class Notes Prepared by: These notes contain copyrighted information Dr. Nina Kajiji Do not quote or copy without permission Last Update: Nov 9, 2009 Trade Weighted Exchange Semi Logarithmic Charts Definition The purpose of the semi-logarithmic chart is to show the rate of change from one period to another. The ruling on the chart is such that the figures are automatically reduced to a percentage basis. Note that the same vertical distance anywhere on the chart shows the same percentage change. Thus, if the interest is in percentage changes in the data, the semi-logarithmic chart is the preferred choice.

7 If two or more series are shown on the same chart, the slope of each line shows the percentage change in the series. By comparing the slopes of the two lines, it is possible to compare the percentage changes in the series. MELec6_6: FORECASTING Page: 7 MBA555 Class Notes Prepared by: These notes contain copyrighted information Dr. Nina Kajiji Do not quote or copy without permission Last Update: Nov 9, 2009 Trade Weighted Exchange Constant Rate of Change Reference Lines It should be noted that a straight line on a semi-logarithmic chart represents a constant rate of change. The primary use of the constant rate of change reference line is to permit the analyst to visually compare the slope of a known rate of change to that of the actual data series.

8 These lines are extremely useful in time series analysis. MELec6_6: FORECASTING Page: 8 MBA555 Class Notes Prepared by: These notes contain copyrighted information Dr. Nina Kajiji Do not quote or copy without permission Last Update: Nov 9, 2009 The Exponential Smoothing Family MELec6_6: FORECASTING Page: 9 MBA555 Class Notes Prepared by: These notes contain copyrighted information Dr. Nina Kajiji Do not quote or copy without permission Last Update: Nov 9, 2009 Moving Averages The moving average approach calculates an average of the sample observations and then employs that average as the forecast for the next period. The number of sample observations included in the calculation of the average is specified at the start of this process.

9 The term MOVING average means that as a new observation becomes available a new average is calculated by dropping the oldest observation in order to include the newest one. Month Period Observed Values 3- Month MA 5- Month MA Log Obs Values Growth Series: Jan 1 Feb 2 Mar 3 Apr 4 May 5 Jun 6 Jul 7 Aug 8 Sep 9 Oct 10 Nov 11 Dec 12 Source: BUSINESS FORECASTING Methods, by Jarrett, (Basil Blackwell, 1991). MELec6_6: FORECASTING Page: 10 MBA555 Class Notes Prepared by: These notes contain copyrighted information Dr. Nina Kajiji Do not quote or copy without permission Last Update: Nov 9, 2009 Monthly Moving Average Sample Date258260262264266268270272123456789101 112 ObservationValuesActual3 Mnth5 Mnth Advantages: 1.

10 Data requirements are small. 2. Better than using a simple arithmetic mean because it can be adjusted to reflect the observable patterns in the data. Disadvantages: 1. The past n sample observations must be available. 2. Equal weights are given to all past observations and no weight is given to observations earlier than period t-n+1. 3. Assumes that the data has a stationary distribution (not always true). MELec6_6: FORECASTING Page: 11 MBA555 Class Notes Prepared by: These notes contain copyrighted information Dr. Nina Kajiji Do not quote or copy without permission Last Update: Nov 9, 2009 Single Exponential Smoothing Single parameter exponential smoothing (Unadjusted) is an easy to implement method of smoothing that overcomes some of the problems associated with moving averages.