Forecasting Case Study: Urban Planning

Forecasting is defined as the process analyzing trends using the past and present data to make certain predictions. During the process of forecasting, it is important for the individual analyzing the data to determine the degree of uncertainty. In other terms, uncertainty is key in forecasting. For this reason, individuals are encouraged to analyze the most recent data to increase accuracy in predicting and minimize their errors ( Box, Jenkins, Reinsel, & Ljung, 2015) . There are various methods of forecasting, but the most common ones include qualitative vs. quantitative, average approach, naïve approach, drift method, seasonal naïve approach, time series methods, econometric forecasting methods, judgmental methods, and artificial intelligence methods. The time series forecasting methods include moving averages, Kalman filtering, exponential smoothing, autoregressive moving average, autoregressive integrated moving average, extrapolation, linear prediction, trend estimation, and growth curve. This paper uses moving averages and exponential forecasting techniques to analyze construction industry data obtained from United States Census Bureau website to determine patterns of economic growth or decline as well as determine the best forecasting method for the next two years. The data is illustrated in the table below and is obtained from New York between 2010 and 2014. The data on paid employees in the construction industry will be analyzed to determine if the industry can be expanded or the resources need to be allocated to a different sector. 

In moving average forecasting method, which is also known as rolling mean/average or moving/running mean/average, the data is analyzed by creating a series of averages. In this case study, two types of moving averages are used; namely, two periods moving average and three periods moving average (Brown, 2004). In two period moving averages, two consecutive numbers are added, and the answer is divided by two to get the average. In three period moving averages, the sum of two consecutive numbers is divided by three to obtain the average. Calculations of averages using 2-point moving average and 3-point moving average are illustrated in the Excel file on worksheet 1 and worksheet two respectively.

Exponential smoothing methods are used to carry out forecasting on data that is changing slowly with time. This is because this forecasting method does not give equal weight to the observations made. In light of this, exponential smoothing gives more weight to the recent observations and less weight to earlier observations in a given set of data (Brown, 2004). 

Among the forecasting methods used in this case study, it can be concluded that exponential smoothing is the most optimal forecasting model. As mentioned earlier, exponential smoothing assigns exponentially decreasing weights to observations as the weight of the involved observations get older. In other terms, the older observations are given little weight while the recent observations are given more weight. On the other hand, all the observations in moving averages are given equal weight regardless of their age. In this consideration, the reason exponential smoothing is a better forecasting model than moving averages for this data is that the error obtained in the former is smaller compared to that realized in the latter. It is important to take note that changes are always occurring, and therefore, giving more weight to the recent is a perfect strategy for enhancing accuracy; thus, reducing the error.

References

Box, G. E., Jenkins, G. M., Reinsel, G. C., & Ljung, G. M. (2015).  Time series analysis: forecasting and control . John Wiley & Sons.  

Brown, R. G. (2004).  Smoothing, forecasting and prediction of discrete time series . Courier Corporation.