Business Forecasting - Essay Example

Assessment 2 Business Forecasting Executive Summary Forecasting is a method which supports several sciences and s to predict the future on thebasis of knowledge of past. These techniques are utilized in the several fields of management from marketing and sales to the inventory control to avoid unwanted hassles and manage the overall activity in a predefine manner. This enables control and management of resources better than doing in any other way. The analysis is done to analyze and identify the appropriate model of forecasting for the softwood unfilled orders forecasting. The number of unfilled orders represents the accumulated inventory. If the data is from the point of view of sales of the softwood supplier it indicates positive. The number of unfilled orders according to the data is going down. The analysis of the data, the methods which have been tried out and the reason behind trying these methods is given below. The papers also support the MS EXCEL sheet which has been prepared in order to analyze the data and forecast. The appropriate method is described below. The basis of choosing that particular method over the others is mentioned at the end of the paper. Accompany Final Model and Forecasts Data Analysis Data analysis is done in various ways. The data when a manager looks at it he might look for the major variations, drops or growths, units time line etc in the first look. The data is useful or not it is decided here. The second step is presenting data in a graph or chart which gives clearer picture of the changes over a period of time. The number of data and the variables present are the basis to analyze it for the further forecasting process. Following the same steps the data provided have the following features in it. The data provided are of 110 months. This is about the monthly unfilled orders of the softwood industry. It’s a price elastic product with high substitute value. If we look at the graph plotted of the original data, we can see a downward slope of the line. The price of unfilled orders is over the period of time is gradually decreasing on an average level. This is an indication of a trend over a period of time in 110 observations which we are able to see. There are fluctuations usually on a difference of every 12-13 months. The price of unfilled orders is going to the maximum of that period during that. This can be due to some external economic factors. When Data is measured over time, observations in different time periods are frequently related or correlated. This correlation is measured using the auto correlated coefficient. (Hanke et al, Business Forecasting, Page 56) Graph of the Unfilled Order over 110 Months The data available has various factors in it which has been deciding factors of adopting a forecast method. The available when analyzed with Autocorrelation method, it has shown the following features in it. .r 1 = 0. 807 r 1 = 0. 683 As the positive lag 1 and lag 2 show that the monthly unfilled orders are related to each other. The value of lag is decreasing, as the number of time lags increases the value of lag decreases. As mentioned in by Hanke et al in their book Business Forecasting (pg 60): If a series has trend, Yt and Yt-1 are highly correlated. The auto coefficient lag 1 is often very large (close to 1). As we can see from the above analysis the value of r1 is 0.807 which is close to one. Not only this if have we observed the Graph of the unfilled orders over a period of time we find that it is decreasing. It can be inferred that the overall unfilled orders is decreasing over a period of time in the 110 observations. It shows a trend of decrease which is a positive impact for the organization or the industry. There are other factors which can have impact on this industry. These are various social, economic, technological and political factors which have been assumed to be constant over the period of time. Methods which can be used here are as follows: As mentioned in the book referred and the class notes that the forecasting techniques to be used in the data with a trend can be as follows: Moving Averages Exponential Smoothing Simple Regression Growth Curves Exponential Models ARIMA Moving Average The moving average is taken for a small time horizon for the data with seasonality, trend or stationary data. The minimum data required for it is 1. It is based on the concept of averages. The averages taken here are of the most recent period. Smaller the numbers of data taken in the averages greater will be the weight given to the data of recent periods. Moving average can be simple moving average and double moving average. The Values of several Errors are as follows: The drawback with simple moving average is that it will not be able to give forecast for more than one period ahead. It depends upon the past data. This draw back led to rejecting this model for the analysis. The requirement for the analysis is the forecast for next 4 consecutive periods which is not possible with it. Mean Error -13.35234742 Mean Absolute Deviation 68.38485915 Mean Squared Error 8869.756098 Mean Percentage Error -0.29% Exponential Smoothing This takes account of most recent observations. It is also based on the averaging only. This method is based on averaging past values of a series in a decreasing (exponential) manner. The observations are weighted and the more weight is given to the most recent observations. (Pg 107) Exponential smoothing is a procedure for continually revising a forecast in the light of more recent experience. (Hanke et al, Business Forecasting, pg 108) Formula used in the analysis Forecast Y (t+1) = 0.7Yt-1+0.3 (Forecast for Yt) The chart developed from this analysis is as follows: The actual and forecasted data lines are visible in the above graph. This graph presents the trend which was observed in the initial observations of the graph and data. Mean Error -99.497175 Mean Absolute Deviation 744.81346 Mean Squared Error 1917943.8 Mean Percentage Error -0.0188073 Time Series -Regression Analysis: We do time series analysis when a set of observations taken at regular intervals are available to us. Time is the independent variable here. Assumptions: What has occurred in the past will occur in the future. We have understood various components of time series analysis which are trend, seasonal variation, cyclic variation and random variation. The graph of the data given shows the presence of trend. For the analysis of data with trend we follow techniques like regression analysis, moving average, centered trend etc. There are two kind of regression; simple regression and multiple regressions. It is done with a data of intermediate or long time horizon. The output from the excel sheet ANOVA (One Factor) analysis can be analyzed as below: 1. Correlation: It is -15.207, which indicates the relationship between X and Y , time and unfilled orders respectively. 2. Regression Coefficients: This will give us the equation of regression. Intercept: 6747.35 Time Coefficient -15.207 Y^= 6767.35-15.207 X 3. Standard Error of regression Coefficient is 2.16. This value is standard deviation of the sampling distribution of the regression coefficient value. 4. Computed t value: 48.789 5. Std. Error of Estimate: 138.29 which indicates that Y value falls typically about 138.29 units of the regression line. 6. P value: It is very small and negligible. It is concluded that regression slope coefficient is significant. 7. R suared: 0.3140 indicates that 31.4% of unfilled order variance. 8. Multiple R: 0.560 9. Significance F Value: It is a very small value here. The F value, if is large than it can reject the hypothesis suggesting significant regression. 10. Adjusted r squared: 0.307, r-square is adjusted for the appropriate degree of freedom. 11. Sum of Squared error (residuals): The sum of squared error is the difference between the actual Y (unfilled orders) and the predicated unfilled orders Y^. Mean Error 1.94301E-12 Mean Absolute Deviation 575.6666042 Mean Squared Error 509373.951 Mean Percentage Error -0.014615663 Justify Choice of Forecast Methods: The best forecasting technique will be the Time Series Analysis with Regression Model, which takes data of the most recent observations as available with us at the same time have an equation which will help us calculating the forecast for the next coming periods. Y^= 6767.35-15.207 X With the help of this method we will be able to forecast for the next four periods more accurately. Y^=6767.35-15.207 X For Month 111 to Month 114 the Unfilled Order Forecast: The forecast for the unfilled orders for the next four moths on the basis of regression model in Time series analysis would be : Y (111) = 6767.35-15.207 (111) =5059.37 Y (112) = 6767.35-15.207 (112) =5044.16 Y (113) = 6767.35-15.207 (113) =5028.95 Y (114) = 6767.35-15.207 (114) =5013.75 As we see from the above data its showing consistent decrease, which is result of the overall trend over the period of 110 months. The forecast method in which the values of MAD, MAPE, MSE is lowest we should chose that. It should not have bias in the forecast. Reasons of Rejecting other Models The reason of rejecting other models over this is as follows: Offers more reliable forecast More appropriate for Inventory control. Supported by the auto correlated study. More reliable than the other two models. Offers better variety of time dependent data. The values of errors like are low in the chosen model. Conclusion On the basis of the given data we analyzed that the trend of unfilled orders are going downwards. That means the unfilled orders will be lesser in the future. The forecasting techniques support the managers and decision makers to take decision on the basis of analysis and knowledge. The qualitative and quantitative methods are utilized in several decision making processes. There are various factors which have impact on and kind of future predictions or forecasting. These factors vary from the social, technological, political, legal or economic factors to the natural factors. Few of these can be incorporated in the analysis but the natural factors are difficult to adjust on advance basis. References Hanke, John E., Dean W. Wichern, Arthur G. Reitsch, Business Forecasting, 7 Edition, Pearson Education, Inc, 2003, Read More