This preview shows pages 1–3. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: Executive Summary There are several procedures to forecast the monthly retail turnover from 2009 onwards, so from April 1982 to December 2008 we have analysed the historical monthly turnover of the business Vegemart and found interesting features. A number of important features of the series are apparent. The features states that the retail turnover of each year increases to its maximum turnover point during December and this turnover drops near June, this is depicted as seasonality. The volatility of the seasonal fluctuations does not grow over time; in fact it stays quite stable as the level of the series increases. Therefore, an adoption of seasonal dummies is required in forecasting. As the years went on, we can see that there is upward movement of the retail turnover of each year increases with time and this is depicted as trend. This trend appears to be nonlinear so we adopted a quadratic trend model (in logs). We notice the historical retail turnover is different in each month and that it sometimes seems to fluctuating a lot or having an up-and-down pattern, this is referred to the cycle component of the series. Assuming if the main historical features remain unchanged, we expect the same pattern or seasonality to occur in the year 2009. In addition, since the trend has always being trending upwards, we might as well expect the retail turnover to be higher than those in 2008. As we said earlier, an adoption of the seasonal dummies and quadratic trend is required to confirm if these expectations are true. Finally we need to consider cycle component of the model, we would expect in the year 2009, the retail turnover will not be the same across months, in fact they might fluctuate a lot or less than the previous years. To predict the future values of 2009 and beyond, we need to construct a model that involves variables that depict trend, seasonality and cycle. The variables that are needed is quadratic trend, seasonal dummies, and ARMA disturbances. To confirm if out model is good or not want the predicted values to be close to the actual values and that these actual values does not go outside the forecast interval. Main Report 1. Introduction In order to create a model using the past data from April 1982 to December 2008 we need to identify the datas trend, seasonality and its cycle. In this section, we will discuss these parts and identify why it is essential that these parts should be included in our potential forecasting model. The forecasting model will forecast the period from January 2009 till January 2010. Below is a graph that depicts the turnover of each year from 1982 to 2008. This graph is in fact shows the log of turnover because there are a number of arguments for using the logarithm (rather than the level) of a non-negative time series. Firstly, when the variable i.e....
View Full Document
This note was uploaded on 03/05/2012 for the course ACCT 2542 taught by Professor Knapp during the Three '11 term at University of New South Wales.
- Three '11