Question

Different types of time-series forecasting models and their applicability in different organizations are given below:

1.       Naive approach: In naive approach, demand for the next period is assumed to be same in the most recent period. This method can be used in economic and financial time series analysis. It can be used to forecast demand for mature products having level or seasonal demand without a trend.

2.       Moving average: This method uses a number of historical data to determine the forecasted value. This method should be used to predict the demand of industry containing mature products.

3.       Exponential smoothing: It is a weighted moving average forecasting method. This method is suitable to determine forecast for most of the industries.

Please give an example of each. 1-3

Answer 1

Given the data for the sales for given months as below,

1) Naive Approach: In this approach, the Forecast for next month is the actual sales of the previous month.

Forecast_Sep = Actual_Aug = 22

2) Moving Average: In this method, the average of the last given (five) months’ data will be taken as next month’s forecast.

Forecast_Sep = (Actual_Apr + Actual_May + Actual_June + Actual_Jul + Actual_Aug) / 5

= (22 + 21 + 18 + 28 + 22) / 5 = 22.20

3.) Exponential Smoothing Method: In the exponential smoothing method, the forecast is computed using the below formula

Forecast_Next = Forecast_Last + [Alpha x (Actual_Last – Forecast_Last)]

Assume that, Forecast_Mar = 18.00

Forecast_Apr = Forecast_Mar + [Alpha x (Actual_Mar – Forecast_Mar)]

= 18 + [0.30 x (16 - 18)] = 17.40

Forecast_May = Forecast_Apr + [Alpha x (Actual_Apr – Forecast_Apr)]

= 17.40 + [0.30 x (12 – 17.40)] = 15.78