Time Series Forcasting - Coursework Example

TIME SERIES FORECASTING INTRODUCTION: Businesses have to make adjustments according to the constantly changing environment. These adjustments require good abilities of predictions about what would happen in future. Capability to meet the market requirements through commendable decisions depends upon the standard of forecasting (Johnston, 2014). Firms that have accurate forecasting always have competitive advantage over those that do not have appropriate or compatible forecasting techniques. Forecasting is a crucial element in the responsibilities of manager and helps managers to accomplish the goals of an organization. Categories of Forecasting: There are two categories in which methods of fore casting fall: 1. Judgmental or Qualitative Methods: Qualitative methods are based judgments, intuition, opinions and past experiences. 2. Quantitative or Mathematical Methods: Quantitative methods are based on mathematical calculations and the traditional types are regression and time series. There are many methods (time series, smoothing, and regression method) available for forecasting and their significance vary according to the situation. It depends on the situation to use the forecasting technique. There is no one method that is “fit for all”. Time series forecasting is one of the best methods; used for forecasting. It is the simplest method and can provide accurate forecasting especially for small time duration. Business in which forecasting of a year or two is required “time series method” is most suitable for them. The forecasting of “Time series” method is solely based on secondary or historical data (Mason, 2006) and forecast the market behavior. The time series forecasting method considers the collection of data that is documented over the time period such as daily, weekly, monthly, quarterly and yearly data. The two examples of time series method are Microsoft Corporation and Sulphuric Acid. Time series forecasting method is used to make plans and take decisions. Long-term forecasting using the time series data can be extended from one year to 20 or more. All techniques that are used in time series analysis are endogenous. Components of Time Series forecasting: Four components that are associated with time series forecasting are: trend component, cyclical component, seasonal component and irregular component. 1. Trend Component: The measurement of regular intervals is taken in time series analysis that is based on hours, day, week, months, and years. It is used to show the random fluctuations. The gradual shifts in time series is known as trend. It is observed that the shifting occur due to the long term factors such as change in technology, consumer preferences, change in the population’s characteristics. For example: A car manufacturer analyzed the monthly variability in the sale of cars. However, when he observed the sale of cars over the past 15 to 17 year, he identified that there is an increase in sale of cars. Suppose in 2005; manufacturer sold 20,000 cars, in 2006 he sold 30,000 cars, in 2007 he sold 34,000 and this increase lead to the sale of 70,000 cars in 2012. However, he identified variability in the sale’s volume when observation was based on months. SALE OF CARS Figure 1 The above figure is an example of trend in time series forecasting method. The upward line is showing the trend that sale of cars id increasing year by year. There are three more types of trends that are possible in time series such as non linear trend, linear decreasing trend and no trend. Figure 2 in the appendix, shows that initially there was an increase in car’s sale then sales of the cards decreased. Figure 3 in the appendix shows that there is a decline in the sale of cars and should not be produced more. If manufacturer wants to produce more cars he needs to consider related issues. Figure 4 in the appendix presents that there is no increase or decrease in sale. 2. Cyclical Component: Cyclical component is another part of time series forecasting method. This method is a chronic progression of point that is above and below the trend line; fixed more than a year known as cyclical variation in time series forecasting. These movements occur due to the cyclical movements in economy that occur in multiple years. This cyclical method is used by many productions. If the general business cycle is observed, then it can be identified that it is based on the period of depression, recession and recovery. A continuous fluctuation can be easily observed through this method. Figure 5: Cyclical Variation 3. Seasonal Component: This method is used by those manufacturers and suppliers when the sales are fluctuating and these fluctuations appear with a seasonal change. Therefore, the component that shows variability in data because of the impact of season is called seasonal component in time series data. 4. Irregular Method: It is a residual factor and includes the deviation of actual time series. These deviations are from those that are expected to give the effect of other three components. Irregular method explains the casual variability in time series. Non-recurring, unanticipated and short-term factors are reasons of irregular components (Hamilton, 1994) APPLICATIONS AND OBJECTIVES OF TIME SERIES FORECASTING: Time series forecasting method can be applied to evaluate the alternatives of different economic strategies, for the evaluation of the model, budgeting, manufacturing and capability development, for the control of inventory, for sale forecasting, for financial risk management and economic planning (Chris, 2000). Time Series is a descriptive technique and the objective of time series forecasting method is to describe the data through the usage of summary statistics, estimate the value of future, explaining the variations and controlling the extra activities. Through this forecasting technique, managers identify the requirements and make decisions for future. Time series method is a reliable method and best for somewhat stable situations. This method is not suitable for such situations in which change is extremely high. Before starting the forecasting; analysts have to complete the collection of desired data. For example analyst must know that how much data is required for the forecasting then he must develop the hypothesis. The next step is the selection of appropriate forecasting model, after the completion of these steps the analyst is ready to do forecast. All techniques of time series analysis can be divided into two groups open and fixed model techniques of time series. Through the OMTS technique, the type of pattern that exists is analyzed and through this pattern a unique model is built for the projection of future pattern. The FMTS technique follows the fixed equation; bases on historical data for the projection of future patterns. FTMS is an easy and economical technique and demand smaller data storage. There are many reasons due to which the time series method is use and from these few are: desire to the deviation among variables, have an understanding the mechanism of data generating, prediction of future time series values and allowing the control and the monitoring of the performance of the system finally. This test enhances the ability of studying the situation, their components and effects. It is a powerful technique for the study of the processes that are used in business. Through this technique lagged effects’ hypothesis can be tested; when two or more bound variables are used in the study. Time series analysis has an advantage of reproducing the non-linear and linear relationship among variables. In contrast, the repeated measurement of participants can influence the perceptions and behaviors of them. Analyzing the group’s effect from sequential trend is not easy over time (Harvey, 1990). One more technique called smoothing is used in time series method and through the usage of this method analysts produce the horizontal data for forecasting. Smoothing method reduces the irregularities from the available data and discloses the real facts. It also removes the seasonal effects from the data. There are two types of smoothing method. One is moving average method and the second is exponential smoothing method. Another method is descriptive method; this method is used to estimate the standard deviation, correlation and mean. In order to describe the continuous effect of past in future autocorrelation function is used in descriptive method. LIMITATIONS OF TIME SERIES METHOD: It is analyzed that time series analysis are based on historical data and historical data does not always give the best forecast. Missing data is another challenge in time series method. The data collection method should be same otherwise it will damage the results. This method is not suitable for such situations in which change is extremely high. CONCLUSION: From various studies it can be analyzed that the use of time series data is extremely important for forecasting purpose and this technique can be applied in diverse situations. It is an important and growing area of research, and provides huge scope for work. Particular and historical time period is used in time series technique. The major aim of this technique is to provide the better forecast in order to make the decision according to the present situation. Diverse methods such as irregular, trend, seasonal, cyclical, descriptive and smoothing have been observed in the paper. It has been observed that there is no such method that is good for all situations. The accuracy of method depends upon the situation. As it has been analyzed that there are many methods - like time series, smoothing, and regression method - available for forecasting and their implication differ with diverse situation. Different forecasting methods work best in different circumstances. There is no one method that is “fit for all”. Recently Analysts and institutions are constantly doing efforts in order to enhance the capabilities of forecasting through workshops by bringing international researchers on one platform. It has been realized that recently semi-parametric and non-parametric techniques are added in time series analysis. Through these techniques the data can be explored in many different perspectives. In Singapore; there is a bunch of expertise that exists in the methodology of non-linear time series analysis. Application of this technique is in environmental effects on the health of public, infectious diseases and finance. Now the data sharing and gathering has become easy due to the technological advancement. The data can be acquired easily and easy availability of data makes the use of time series method easy and increases the chances of the use of time series method in different organizations (Brockwell, & Davis, 2002). References Brockwell, P. J., & Davis, R. A. (Eds.). (2002). Introduction to time series and forecasting (Vol. 1). Taylor & Francis. Chatfield, C. (2000). Time-series forecasting. CRC Press. Hamilton, J. D. (1994). Time series analysis (Vol. 2). Princeton: Princeton university press. Harvey, A. C. (1990). Forecasting, structural time series models and the Kalman filter. Cambridge university press. Johnston, K. (2014). What Is the Relative Importance of Forecasting?. Small Business, Retrieved March 1, 2014 from http://smallbusiness.chron.com/relative-importance-forecasting-35627.html Mason, N. (2006). Forecasting Techniques, Part 1: Quantitative Methods. Clickz, Retrieved March 1, 2014 from http://www.clickz.com/clickz/column/1695851/forecasting-techniques-part-quantitative-methods Appendix Figure 2 Figure 3 Figure 4 Read More