The focus of the first group assignment is to offer a learning opportunity to each group member to get familiar with the basics of forecasting, before undertaking individual projects on their own. It must be mentioned that forecasting is as much an art as managerial science, and is best learnt through ‘hands-on’ experiential learning through practice. The prerequisite course FSMA-6010 helped us learn both univariate and multivariate econometric modelling. In this course, we will learn to forecast without using explanatory independent variables, and if auto-regression exists, the past time period is the best predictor.

Requirements:

Refer to the following data website: https://www.census.gov/retail/marts/www/timeseries.html

Steps:

1) Your group needs to choose and select one time series from the list of monthly time series data.

2) You will need to partition the data series, selecting only part of the data to model, leaving say most recent 12 values (monthly data series) or 30 values (daily data series) to test the accuracy of your forecasts.

3) For this assignment, we need to use Exponential Smoothing. You may use Excel Data Analysis ToolPak (In Excel, go to File ---> Options --->  Add-Ins--->  ToolPak---> Select Analysis ToolPak

Refer: https://www.statology.org/how-to-load-the-analysis-toolpak-in-excel/

4) Use only part of the data series to finalize your forecast model. You need to forecast one-period ahead, incorporate the forecasted value and to forecast the next one-period ahead. To generate a series of forecasted values.

5) As you finalize your model dependent on the chosen time series data, you will need to analyze with respect to ‘Trend’ and ‘Seasonality’ using Holt-Winter’s Method (Refer Handout).

6) Use accuracy measures such as MAPE and RMSPE (refer attached handout) to analyze the accuracy of your forecasts for the time series chosen.

7) Submit your analysis as a Word document, incorporating summary tables.