qo) INTRODUCTION The time series is an arrangement of statistical data in accordance with the time of its occurance. Time series refers to any statistical data collected at regular intervals of time. Time may be in terms of years, months, weeks, days and so on. Mathematically, a time series may be defined as y =f (#), where y is the value of the variable at time f. Thus, a time series gives relationship between two variables-one is independent variable time (t) and the other, dependent variable (y). For example the price of a commodity (y) in different years (t), the production (y) of tea in Assam in different years (f), the sales y) of a store in different months (f) etc. 10.2 IMPORTANCE OF TIME-SERIES ANALYSIS yusinessmen, economists, The analysis of Time Series is important to b ious other fields due to social scientists and also to people working in vari the following reasons. ( It helps in understanding the past (ii) It enables to isolate the various forces affecting the time series. (i) Analysis of time series helps in forecasting the future value of a variable. (i) A time series provides a scientific bi between two or more sets of data. (v) Time series helps in planning future operations. behaviour of the variable. asis for making comparison4 ‘Irst Course in Statistice-1 [02] % AF "70.3 COMPONENTS OF TIME SERIES Th changes in the values of a variable with respect to time can be the resultant effect of a variety of factors, These factors can be classified into four categories which are called the components of time series. Some or all of the components may be present in with varying degrees. The components are as follows: (a) Secular Trend or Trend (b) Seasonal Vi (©) Cyclical Variati any time series ations (@) Random or Irregular Variatior m Trend The trend is the general long term movement in the time series value of the variable over a fairly long period of time. The trend of a series is generally either upward or downward in nature. For instance, the data relating to population, production, literacy ete have upward trend while the data relating to death rate, illiteracy, epidemic etc. have downward trend, However, such a trend may not always hold good. It is to be noted that depending on the nature of increase and decrease, trend may be linear as well as non-linear. Again long, period of time can not be defined exactly and it would depend on the nature of the problem. im Uses The study of trend helps a businessman in forecasting and planning his future business activities. It helps an economist in formulating economic policies. It is used in making comparison between two or more time series. By isolating trend from the given time series we can study the short-term and irregular variations. @ Seasonal Variations Seasonal variations refer those pattern of change in a time series that repeat over a period of one year or less and they repeat from year to year. They can be studied when the data are recorded at half yearly, quarterly, monthly, weekly etc. They cannot be studied when the data are recorded annually. Seasonal variations occur regularly season after season and therefore they are definite and can be foreseen with a little effort. They are due to the following two reasons. ( Natural Cause : Changes in the climate and weather conditions have a profound effect on seasonal variations. For example, the sales of umbrella pick up very fast in rainy season, the demand for cold drinks goes up during summer etc. Man-made Cause : The social customs, traditions and conventions play an important role in fixing seasonal variations. For example, during, festivals like Bihu, Puja, Id etc. the sales of many commodities go high, the price of ornaments go high during marriage seasons etc. Uses An accurate assessment of seasonal variations is an aid in business planning and scheduling such as in the area of production, inventory control, personnel, advertising and so on. A knowledge of seasonal variation helps a consumer in purchasing the articles at least prices during the slack seasons.Recession noone ry Depression Business Cycle Fig. 101 Renn! | mCyclical Variations The cyclical variations refer to the movements which occur after time intervals of more than one year. These are the changes that take place due to economic booms ay depressions. The cyclical variations, though more or less regular are not uniformly periodic. One complete period normally mean 7 to 9 years. Series relating to price, production, wage etc. are all affected by ‘business cycles’. Business cycle is composed of four phases, namely-boom, recession, depression and recovery. Each phase change gradually into the next phase Uses The knowledge of cyclical variations is of great importance to businessmen in the formulation of policies for stabilizing the business activity. With a knowledge of cyclical variation a businessman can face the challenges of recession and depression by taking the appropriate decisions in advance. | Random or Irregular Variations ‘As the name suggests, these variations are random, accidental or simply due to chance factors. These variations may be due to such isolated incidents as floods, famines, wars, earthquakes etc. which are completely irregular in nature. Thus, they are wholly unpredictable. ‘This includes all types of variations in a time series which are not attributed to trend, seasonal or cyclical variations. 10.4 MODELS FOR TIME SERIES ANALYSIS: (i Additive Model: This model assumes that all the component of the time series are independent of one another and each part can be measured independently. Here y=T+S+C+l Where Y is the time series value and T,C,S and | are Trend, Seasonal, Cyclical and Irregular variations. (i) Multiplicative Model : This model assumes that the four components of the time series are due to different causes but they are not necessarily independent. The traditional time series analysis model is characterised by the multiplicative relationship, y=TxSxCxlr ics-1 A First Course in Statisti > % fm Moving Average Method i calculating a series of arithmetic means of som hod of moving average consists in calculating me . eae . There is no hard and fast rule for decision regarding lection of period depends upon the objective that has to es ups and downs in the overlapping groups of the time seri period of moving average and the be attained from the figures of trend. The averaging process remot are the values of the time series for time periods ty, ty, 5... then data. If vp ¥z ¥: SH (ty Hot Yop) Ist moving average of period m mn 1 WO? + Y3 4+ Ying) 2nd moving average of period m _1 atm Os +94 +--+ Yma2) ete. Srd moving average of period m Now, if the period m is odd, then the moving average values are placed against the middle values of the time periods. Again if the period m is even, then the moving average values are placed in between two middle time periods. After that, method of centering is applied to adjust the data with the original time periods. The moving average values plotted against time give the trend line. @ Merits and Demerits of Moving Average Method Merits () This method is easy to understand and calculate. (i) This method is not subjective. (ii) This method is flexible in the sense that a few more observations may be included without affecting the previous results. (@) If the period of moving average is equal to or multiple of the cycle of a cyclical variation, then the effect of cyclical variations can be removed. Demerits () Moving average can not be obtained for all the years. ii) There is no definite rule for fixing the period of moving average. (ii) This method is suitable only when the trend is linear. (iv) This method is not suitable for forecasting.oe Frample-4. Calculate 5 yearly moving average from the following data: Year : 4995 | 1996 | 1997 | 1998 | 1999 | 2000 | 2001 | 2002 | 2003 | 2004 Production : 52 | 79 | 76 | 66 | 68 | 93 | 87 | 79 | 90 (thousand tonnes) Solution. Year Production 5 yearly moving 5 yearly moving average (000 tonnes total 1995 52 - 1996 79 - 1997 76 341 68.2 1998 66 382 76.4 1999 68 390 78.0 2000 93 393 78.6 2001 87 417 83.4 2002 79 444 88.8 2008 20 2 2004 95 _ re >expressed by a mathematical equation. Example-g, — Using least squares method find the trend values from the following data: Year : 1990 | 1991 | 1992 | 1993 | 1994 | 1995 | 1996 Production = 83 60 54 21 2 13 23 Solution. Let the equation of the straight line trend be Since n is odd, y =a+bt _ year - 1993 i 1 = year - 1993~ A First Course in’g tatistigg [032] — ‘Trend in Estimateg 5 ‘Year, Production Production Y=3943-10.93, 72.22 : 83 i ns 61.29 tet = 50.36 1993 a1 au 1994 22 iS sees = 1757 1996 23 = | S276 b —1093 a er Thus, the equation of the Straight line trend becomes ¥ = 39.43 - 10.93 ¢ Example-1o, —— [Year 1980 | 1981 | 1982 | 1983 | —teaa 1985 | 1986 | i987 [Sales = 12 13 13 16 19 23 2i 23 Solution. Let the equation of the straight line trend be ¥ =a+ bt Since n is even, t= ear — 19835 eS (year - 1983.5) 2 nd in Estimated, Sales y=17.510.01t 17.43 17.45 17.47 17.49 17.51 17.53 17.55 17.57 (year ~1983,5) cf, 56 3 -1 1 3 5 1Zyt ze es ~ 168 =. The equation of the straight line trend becomes y =175+ 89t Now, for the year 1988, the value of f is 9, and hence the estimated value is y =175 +89 x9 25.51 a - NOTE: When 11 is odd and time interval is 1, b can be taken as yearly increment and when nis even and time interval is 1, b can be taken as half yearly increment. 89