— — ——— 16 ANALYSIS OF TIME SERIES pers a ile dae eae a Timefactoris a Continuous dynamic affair, and the sequence of fluctuations chron Sconomie, business, Political, social and various other areas with such Bical movement, These fluctuations affect various economic aspects ay 8 trade, Commerce, industry, employment, price, ete. Hence, the study and this nn. Movements linked with time factor have become very useful and for * Purpose the technique of analysis of timo series te used. . MEANING OF TIME SERIES Time series tefers to such a series in which statistical data are presented on the basis of time of Occurrence or in a chronological order. The measurement oftime may be either Year, month, week, day, hour or even minutes or seconds, cally, in a time series the time factor is an independent variable and the Gependent variable in the form of statistical data represents the movements Telated to time factor. The following two series are the examples of time series : Population in India Wholesale Price Index No. (2004-05 = 100) Year Population (Crores) | Month (Year 2012) | Index No. 1971 54.0 April 197 1981 68.4 May 195 1991 84.4 June 196 2001 102.7 July 196 2011 120.2 It is worth mentioning here, that the time series is also termed as ‘Historical Series’ or ‘Chronological Series’, A few definitions of time series are as follows : 1, “Asset of data depending on the time is called a time series,” —Kenny & Kieping 2, “A time series is a sequence of values of the same Variate corresponding to successive points in time”, —Werner Z. Hirsch 3. “A time series may be defined as a collection of : ‘ reading belonging to different time periods, of some economic variable or composite of: variables”,—<—<—$ ANALY teas F TIME SERIES 579 ap ae sie paren poten is extreme! . z sh ‘ mete =r the following polite me ly useful in business decision-making it ysis of Past Behaviour : Analysis of ti . A . 8° Analy var of a variable wh aoe Penne es bere in este ee st cent Cf iscloses the circumstances and causes {Ping the movement of data. This information can be helpful in the ent for ge behaviour of the variable. o Forecasting ° The Ae jowledge of trend and fluctuations on the basis of ., of time series makes it possible to forecast the future behaviour of the #5" snder study. On this basis businessmen, administrators and planners Oy down future policies and targets and can prepare future plan. Prof. ook Hirsch See The main objective in analysing time series is to Weptonts interpret a evaluate changes in economic phenomena in the hope nore correctly anticipating the course of the events.” Comparative Study : Analysis of time series facilitates the comparative oily of data for two or more related periods. For example, to compare the data of production in a factory during a period of ten years from 2000 2010. 3 v 4, Evaluation of Performance : It helps in the evaluation of current i ts. For example, the progress of long-term plan can be evaluated on ihe basis of comparison of yearly targets of growth rate and actual annual ie ts. a - os . 5, Estimation of Trade Cycles : Trade cycles can be estimated on the basis ofqyclical fluctuations and it helps the businessman to plan and regulate his | activities. COMPONENTS OF A TIME SERIES Time series is influenced collectively by a large variety of factors and forces. ‘The effects of these forces can be classified in some definite categories. These categories are called the components of time series. The main components of time series may be classified as shown below : Original Data (0 or¥) Long-term Movement Short- time Fluctuations or (OD) Trend or Secular Trend Regular Short-ime Irregular or ® Fluctuations (S +C) Random Fluctuations @ Sasonal Variations (S) Cyolicsl Fluctuations (C) These various components of time series have been examined in detail in te ftllowing paragraphs : LSecular Trend or Long-term Movement or Trend Trend refers to that tendency which indicates the general direction of that inalong pero. Insimplo word isan estate that dente various tions from time to time, there will be an underlying tendency of in a particular direction and this tendency is called as long-term: a 580 BUSINESS STATISTICS trend. For example, that despite the fluctuations in prices in our eoy; long-term trend is of increasing. There are certain such facts also, . tN, the tendency moves to one direction only such as continuous increase in pc st hic continuous decline in death rate, etc. Populat Some important characteristics of secular trend are as follows : : (a) Three Aspects : There may be three aspects of long-term trend: Yj, | trend, Downward trend, and Stable trend. An upward trend is usually obge at in time series relating to prices, population, etc., while a downward Ted noticed in data of illiteracy, death rate, ete. The stable trend is foung ; is and natural sciences such as tendency of temperature of an individual, aibege (b) Different Trends during Different Periods : It is not necessary dae trend should be in the same direction throughout the given period. It the possible that different tendencies of increase, decrease or stability are o| a in different periods of time. However, the overall tendency may be downward or stable. at (©) Relative Concept : The term ‘long period of time’ is a relative which is influenced by the characteristics of the series. For example, during thy days of curfew the period of one month may be a long period in the context of data relating to number of deaths, while in price statistics a period of seyeray years will be considered as a long period of time. The symbol of “7” is used for denoting long-term trend in the formulae relating to analysis of time series. 2. Regular Short-time Oscillations Most of the time series are influenced by such factors or forces which repeat themselves periodically, The variations arising out on account of such regular or periodical repetitions are called regular short-time oscillations, which may be classified into the following two categories : (a) Seasonal Variations : Seasonal variations refer to such movements which are regular and repetitive and which operate in a regular and periodic manner over a span of less than a year. This span may be a day, week, month, quarter, half-year, etc. Characteristics : The main characteristics of seasonal variations are a3 follows : (i) Regular movement : Seasonal variations occur regularly almost at the same time and about the same proportion within a period of less than one yea" (ii) Swings in both directions : These variations may swing 1? ay direction—upward or downward. , (iii) Easy forecast : Seasonal variations can easily be forecasted. ‘Various economic and business activities are operated on the basis of these forecasine Consumers, producers and sellers give due consideration to these ee while taking decisions about their operations. These variations are denoted letter ‘S’ in the analysis of time series. ss (b) Cyclical Fluctuations : Cyclical fluctuations also occur Pel ine seasonal variations, but the period of their reoccurrence is more than & ol These fluctuations are called cyclical because they occur in cyclical nature 00, in this cyele there are four stages—{(i) Prosperity, (ii) Recession, (ii) Depres* (iv) Recovery, which can be demonstrated by the following chart : _ANALYSIS OF TIME SERIES 581 , rth-mentioning that there is no definite period of cyclical It is NGeneraly, this period varies from 3 to 10 ear, The concept of jot yctuation can be understood more precisely by the examples that after Po yree years uate ae se bamper ie of mangoes or after every ook ,oduction of sugarcane reaches e peak, etc. fos 320. ne, the cyclical fluctuation is denated by the symbel of ‘C' I or Random Fluctuations $ irregular OF random fluctuations occur accidently in time series. For decline in profits due to break of fire in the factory in a particular year, isto jn production due to sudden strike or scarcity of petroleum products a to war, ete The following facts are important about the characteristics of fluctuations = r oo No Forecasting Se ena are the caus ofsuch forces working , which cannot be predicted or forecastet (i) No Definite Pattern : Irregular fluctuations do not exhibit any definite pattern and there ae regular period or time of their occurence. That is why, are called irregular. (ii) Short-term = Generally, they occur as short-term variations but sometimes their ee may be so intense that they give rise to new cyclical or other movements. (iv) Coverage : Irregular fluctuations cover all such variations in a time series, which are not covered within the gamut of trend, seasonal and cyclical movements. Inthe process of analysis, irregular fluctuations are denoted by the symbol of T. MEASUREMENT OF TREND" OR TREND ANALYSIS the ioe tolleyne are the fourimportant methods which are used in estimating ()Free-hand Curve Method, _(II) Semi-average Method, (1) Moving-average Method, (IV) Method of Least Squares. Le (@ FREE-HAND CURVE METHOD ee ‘method is also called as ‘Graphic Method or ‘Curve fitting by ‘ion’. The procedure for knowing the trend by this method is as follows : 1) First of all, the original values of a time series are plotted on a graph oe a historigram is obtained by joining these points. (2) After this a smooth eurve is drawn through these points keeping in view tendency n of fluctuations in such a way that the curve represents the general Tot the data, ‘terms ‘Secular Trend’ or ‘Long-term Trend’ are also used in this context.a 582 ____ BUSINESS STATISTICS It is evident that by free-hand curve method an effort is ade a dry @ smooth curve, so that seasonal, cyclical and irregular a thee in the original data may be eliminated. Here, it is nd ahey pei dae t diff, Persons are likely to draw different curves from the Stine & ©. Hence, the following points must be kept in mind while drawing an Smog, curve—(1) The curve should be smooth. (2) The number o at oe thetrend curve should be more or less equal to the number of poin! eh - (8) If there is cyclical fluctuation in data, the cycles above the curve and below the ey,,, should be almost equal. Merits | (1) Simple Method : It is the simplest m no need of any mathematical calculation which (2) Flexible Method : It is flexible in the sense both linear and non-linear trends. Demerits ae | (1) Subjective Method : This method is highly subjective because thereig all possibility of ‘drawing different curves by different persons for the same set | of data. In other words, the curve may be influenced by the personal bias ang judgement of the drawer of the curve. (2) Lack of Accuracy : This method is not based on mathematica) calculations. So, there is lack of accuracy in it. lee (8) Danger in Forecasting : Since the curve is subjective, it will be dangerous to use it for forecasting or making predictions. j : The conclusion is that despite being simple and flexible this method is not scientific and much reliable. So in practice, its use is very limited. thod of studying trend, ‘ | saves time and ‘abou tes that it can be used to represent Illustration 1 Show the secular trend from the following data by free-hand curve method: Year : 2007 2008 2009 2010 2011 ©2012 Production@) : 60 68 64 70 2) aaa Solution | GRAPH SHOWING TREND BY FREE - HAND CURVE METHOD 72 € P= 5 F 3 a g oo ORIGINAL DATA 2O07F=H 08 Fe! 09 FE! 40 Et 41] 2012]ANALYSIS OF TIME RASC 683 SB ——- _ (ID SEMI-AVERAGE METH : oD 7 seocedure of finding out trend by semi-average method is as e + veion of time series into two Ki @ pinion ed into two equal eta eee : First of all, the values of 66h non the vales, of first three years will b us le, if values of six years iPce tho remaining three years in the» Ee ee and ” yalveee the number of values is even, th will be part. It is important to OF phat wut if the number of values is aad the a vided exactly into two aa ee Sa i carats oo egal pie examples if values of 11 years are given, then the value of the middle id ast = 6th year will be left out and the series will be divided into two ad on the pasig of values from first year to fifth year and seventh year jjeventh. nears ; i909) Calculation of two averages : Having divided the given series into two al parts, we next compute the arithmetic mean for time-series values for eS half separately. These means are called semi-averages. Alternatively, a may also be calculated. o (3) Plotting of original data on graph paper : After making these tical calculations the graph paper is used and we draw a curve on the Rs of the plotting of the original data on graph paper. @ Plotting of both semi-averages * Both the semi-averages are plotted sspoint against the middle point of the respective time periods covered by each part. For example, there is a period of five years in each a 5 point will be plotted against the median point of first 5 years, ie, year and the second point against the median point of last five years ie., 8th ‘year. (5) astraight line is drawn by joining the two points of Trend line : Finally, semi-averages and this line represents the trend of data. Ilustration 2 “ Determine trend of the following data by applying the method of semi-average and eimate the value for 2014 : Year + 2007 2008, 2009 2010 2011 2012 Production (thousand units) : 40 yaeeawiaas egde 06 G4 Solution so they can be divided into two ee are six years figures jn this question, ; es of 3 years each i.e., 2007 to 2009 Md 2010 to 2012. The trend line willbe on graph paper on the basis of averages ofthese two halves. 3 Yearly semitotals | Semi-averases Year 2007 ; z ipa 2010 2011 1-60 inst the mid-point The semi Tpalf 44 will be plotted agains wake frst none io ‘and the semi-average of the second half 60 similarly be plotted against the year 2011.584 BUSINESS STATISTICS v a Saae SHOWaG TRENDY SEMI AVERAGE METHOD. [ESTIMATE FOR 2002." 7€ PRODUCTION (THOUSAND UNITS) sees a roe pic oo Yusl 1OFe 145.2012 20134 2014 una YEAR! i Illustration 3 Fit a trend by the method of semi-averages to the data given below + 2004 2005 2006 2007 2008 2009 201 0 2011 2012 wn Year Production (% crores) : 9 12 18 14 17 «+18 20 22 Solution Here the number of years is 9. So the middle year (2008) will be ignored and the time series will be divided into two equal parts: first from 2004 to 2007 and the other from 2009 to 2012. Calculation of Semi-averages ‘Sales (Z crores) Z Yearly Semi-totals Fa The middle of the year 2005 and ill be plotted ss Ye 30 be plotted against the middle 12 will i-average 21 ‘SHOWING TREND BY AVERAGE METHOD, GRAPH SEM ~—— ———— eee. oe ANALYSIS OF TIME SERIES 585 | EE average Method here is of This method is easy to understand pple * : : and apply because tl ) Gee tal calculation except finding out arithmetic mean of two parts of Peers. os ow d line is dr. . 3 gH opivity * ‘Trend line is drawn on the basis of two semi-averages. So, pe Obi ivity and certainty in trend line. In other words, any two persons is ot ear0e trend line from a set of figures, wee or future estimates : The trend line can be extended on either side ore obtain past or future estimates. 4s of Semi-average Method yo pine? nd : This method can be used appropriately only when there ) as proximately linear relationship between the plotted points. ii pyfect of Extreme ‘Values : If there are certain extreme values (very low QT), they will influence the value of semi-averages and in such a case pay not represent the values correctly. vst jsevident that semi-average method is more objective in comparison to and curve ‘method, even then, it is not much reliable particularly when fee ore cortain extreme values or lack of linear relationship. a (IM) MOVING-AVERAGE METHOD Moving average method is a simple and flexible device of reducing tions and obtaining trend values with a fair degree of accuracy. It consists p obtaining & series of moving averages (arithmetic means) of successive weriapping groups OF sections of the time series. For example, there are six pares bc de andf and three year’s moving average is to be computed. It ill be done as follows : atb+c b+ctd c+dt+e dtrer+f Cee ee Uae ares The basic question to be decided in this method is that what should be the period of moving average, ie., three yearly, four yearly, five yearly, etc. This ecsion is taken on the basis of size of data and fluctuations therein. From the point of view of calculation of moving averages, the questions can be divided in two categories : (1) when period is odd, and (2) when period is even. (1) Odd Period Moving Averages : It means moving averages of odd period aryears, i.¢., 3,5, 7,9, 11......years. Its procedure can be explained as below on the assumption that three yearly moving averages are to be calculated : (i) First ofall, three yearly moving totals will be obtained. The total of first three years ibe placed against the centre of three years, i.e., second year. (ii) After it, ofnext three years (second, third and fourth) will be placed against third Year, total of succeeding three years (third, fourth and fifth) will be placed fons ng total by 3. tis Important that moving averages will not be obtained for Years in last year in case of 3 yearly moving averages and first two and last two case of 5 yearly moving averages. me 4 ay calculate trend values from the following data ous 6-year and 7-year moving 1 2 3 4 "5 6 7 8 110 «104 98s—«dH.s109's«120 15110 Homage 14 16 16 9 1o | 6 114 iz «180 «127'«122s18s«180 140140. = (2) Even Period Moving Averages : If the moving average is to be on the basis of even period, ie., 2,4, 6 years, then oer are ee eoee centering the moving totals. Suppose, four yearly moving totals are to be calculated, the following procedure would be adopted : @ First of all, four yearly moving totals will be obtained. The first total will be of first four years, the next total of four years excluding first year and this process will be repeated. The first total will be placed between second and third ar, second total between third and fourth year and so on. (ii) After it, these moving totals will be centred. For this purpose two period moving totals will be obtained. (iii) "Two period moving totals will be divided by 8. Illustration 5 From the following data calculate the 4-yearly moving averages and determine the trend values, Plot the original data and the trend on a graph: Year : 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Value 50.0 365 430 445 389 381 326 417 411 338 Solution Calculation of Trend Values by 4-yearly Moving Averages 2 Periods || Moving Averages* Year Moving Tots |” (trend Value) 2003 2004 2005 gee 2006 40.93 2007 39.83 38.18 38.10i ~~ _ ANALYSIS OF TIME SERIES 587 “es GRAPH SHOWING TREND BY [iti MOVING - AVERAGE METHOD [ORGINAL DATA] fos Hits o7 [years =i oa Ed 09 Je of Weights in the Calculation of Moving Average Weights may also be used in computation of moving average. Its objective isthe assign different importance to the values of different years. It can be alained with the help of following example. ‘Iustration 6 Given the numbers 2, 6, 1,5, 3, 7 and 2. Calculate weighted moving average period 3, the weights being 1, 4 and 1 respectively. Solution Calculation of 3-period Weighted Moving Average Weighted Moving Total : : Values eshte Mone of | Weighted Moving Average (2x146x44+1x1)=27 (6x1+1*44+5xD=15 (1x145%44+3x1)=24 (6x 143x447 1)=24 (3x 147%4+42x 1) =33 ets ot Moving Average Method Wt req = ene : This method is simple to understand and apply since it does Qo complex mathematical calculations. Working Chiectivity : This method is objective in the sense that any person thig. a problem with this method will get the same trend values. It is in T than free-hand curve method. ho co DDD (@) Xethagg bility : This method is flexible enough in comparison to both average and least squares. It means that if certain new values Re added X that meat entire calculation will not have to be done again. Its only effect will be some more trend values.i segue Tamas —pusiness STATISTICS (A) Elimination of eetion! fluctuations : If the wed dae coincides with the period of cyclical fluctuations in the date ving ees are automatically eliminated. 1 Such flucs, (5) Measurement of other components of time series : This . 4 Method not only for determining trend values but also for seasonal, crt Slag irregular variations. Demerits : (1) No trend values for some years : ‘An important limitation of th, js that trend values for some years In the beginning and some at pe be obtained. For instance, in three yearly moving average, the tre; vad any yearly moving average trend values for try first and last year and last two years are not obtained. (2) Determination of the period of moving average : Under thi determination of moving average period is also not an easy task and “ Methoy if there are no business cycles in the series. Particulary (3) No use in forecasting : Since trend values are not expressed on the of functional relationship, this method is not helpful in forecast ais predicting the values on the basis of time. ting ang (4) Appropriate only in regular variations ; Generally, this method j considered appropriate only in time series having regular variations. In : words, this method js not appropriate in other cases. + Ta er : Moving averages is better than free-hand curve and (6) Effect of extreme values extreme values. Inspite of these limitations this method semi-average methods. However, this method is appropriate for fitting a tread when :(a)the trend is linear, (b) the cyclical variations are regular both in perial and (c) the purpose of investigation does not involve current and amplitude, OF LEAST SQUARES analysis or forecasting for future. (Iv) METHOD e of the best methods for obtaining tread This method is considered as on! values. Under this method the line of best fit is obtained the assumption of least squares by using algebraic equal i in the form of straight line or parabolic curve. ‘This method is called the of least squares because the sum of squares of the deviations of vari of trend line from original data would be the least as compared to the sums squares of the deviations obtained by using any other line. ‘The determination of trend on the basis of the method of least squares? be categorised into three 1) Fitting a Straight Line Parabolic or Non-linear ‘Trend, and (8) Semi-logarithmic or (1) Fitting a Straight Line ‘Trend For fitting a straight line trend following equation is used : _ Y,=a+bX where, Y, = required trend value X= unit of time are also influenced by Bxponeatial CH on the basis of method of least squaresa SRT f= ANALYSIS OF TIME SERIES ee a i ee Oe | ~ re constants. If the first year is taken as origi Wi 4° a the point of origin (O) and the point dima ete 74% peter che middle year of the time series ig bisa aa ‘eget rae “metic mean of time series, The constant ‘’ i di Sea ig wpe re jg also called growth rate or decline rate hiesaar 4 tap je jine ) for each unit change in time (X), aa in (fe! two equations are used to determine the values of constants Aphig ro aeons xY=Na+b 3X ) a g: fF IXY =aX+b xx? (ii) ions in the above form i be of equations in rm is called the long method of least ghe power’ jf deviations are taken exactly from the middle year of time i of ZX becomes zero and in such a case the above equations may ete follows : ks ¥ a= x, b= x qo vse of summarised version of equations is called the short method of _Ifnothing contrary 18 mentioned in the question and it is possible ystst22"Gations exactly from the middle year, then short method should be pie pecause its calculation procedure is very simple. (@) Least Squares Short Method : The procedure of this method is as wi “a total 6 columns are drawn, which will be for year, value (Y), deviations ot, x’ and trend values (¥Y,) respectively. (2) First of all, time deviations are taken for all other years from the exact niddle or median year ‘and these deviations are shown in the column of X. It gould be checked that sum of deviations is equal to zero or EX = 0. (3)Each deviation is squared up (%?) and by adding these squares we obtain ‘© EXY is obtained by adding multiplication (XY) of values (Y) and (6)'The value of constant ‘a’ is computed by the formula = Here N stands forthe number of years (time units) involved. (6)The value of is found by applying the formula : x. ate values Y, are obtained by applying fhe formula a + bX against aa thot be checked that sum of he original values (EY) must be equal to i nd values (ZY, or rY’) 1 Ei versed line by the method of east squares: ee ie et eee 50 67 5 85 the value for 2011. ,aa BUSINESS STATISTICS Calculation of Trend Line by Least Squares Method Trend Y =a4 hy 62.2+12.7x-Ia995 622+127x-124q5 0 62.2+12.7x0=679 75 62.2+12.7x1=749 6224127 x2=978 Estimate for the year 2011 X = Gn. =a+3b 2.2 +3 x 12.7 = 100.3 Yoou Illustration 8 Fit a straight line trend by the method of least squares to the following data; Year + 2006 2007 2008 2009 2010 2011 2012 Production C000 Qntls) = 80 90. 92 83 94 99 «92. Show the above series and the trend on the graph paper also. Solution Squares Method Calculation of ‘Trend Values By Least Production | Deviation | squares Pon ae Trend Values 7000 Qntls.)| from 2009 XY atbXx=Y, — 240 90+2x-3=84 - 180 90 +2x-2=86 90+2x-1=8 -92——___ANALSIB OF TIME SERIES Ln figures (Y) and trend values (Y,) will be shown on graph v qv? Ps? allow 08 ie erg TERENCE GRAPH SHOWING TREND BY. LEAST SQUARES METHOD 100 =» TREND. |» ORIGINAL DATA PRODUCTION 000 QTLS. O7 Fy 08 i] 09 Fit 10} 4 12 jon 9 posto ig of data given in Illustration 8, te este production figure for the year 2014. 0 Do these figures show a rising or failing trend ? e pe at is expressed by the difference between original values and trend values ? @) ‘What is the slope of straight line trend ? @) Yoous =a+bX =90 + 2x5=100,000 Qntls. i) Figures are showing a rising trend because slope is positive. (ii) The difference between original values and trend values expresses short-term fluctuations. (jy) The slope of straight line trend or b = 2 (000) Quintal. Mustration 10 Fita trend line to the following data by the least square method and estimate the duction for 2009 and 2014 : Year 2004 2006 2008 2010 2012 Production (000 Tons) : 18 21 23 27 16 Y,=0+bX=21+01X6 RS Estimation for 2009 a Ypq9 = 21 + 0.1 x 1 =21.1 (000 Tons) Estimation for 2014 Yoo14= 21 + 0.1 x 6 = 21.6 (000 Tons) Problem of even number of years or time units : or years is even (i.e., 6, 8, 10 and so on), then deat eT ting : centre of two middle years (or time units). For example, ‘fie fg ttt the question, deviations will be taken from the middle of 3rd ed ate § eat year) and they will be ~ 25, - 15,~0.5,40.5,+1.5 and +28 3eGsq make the calculation simple, these deviations may be multiplied ively % in terms of six month values) and then these 4; iby (devi -5,-3,-1,+1,+3and+ 5 respectively. All other procedures rema, ike hy Fate, Mlustration 11 Calculate the trend values by the method of least squares from: 2, and estimate the sales for the year 2018 : he folowing dy Year : 2007 2008 +2009 +2010 Soke(eLekhs) : 120 140 150 170 aa a Solution Year | rakhs (Z) | 2009.5 and AY | Trend Values a+ bx-y, Multiplied by 2 2007 | 120 “5 25 160+7.71x-5=i1g 2008 | 140 -3 9 160+ 7.71x-3= 13681 2009 | 150 -1 1 160 +7,71x-1= 1529 2010 170 +1 et: 160+7.71x1=1677 2011 |. 180 +3 9 160 +7.71 x= 18313 +5 25 160 +7.71.x5= 19855 Y,=160+7.71X Estimate sales for the year 2013 Yopyg = + 6X = 160 + 7.71 x 7 =213.97 @& in Lakhs) (b) Least Squares Long Method : The calculation process in this method’ as follows: = (1) First of all, deviations are taken assuming any year or unit of time origin. There is no restriction in respect of origin in this method. If see instruction is given in the question, that should be followed, calculations can e simplified by taking mid-point in time as origin. ve rx to avoid negative signa, the points of time or years (X) are denoted by a numbers like 1, 2, 3, 4, .« which means that the year or unit of time J preceding to first year or unit of time has been assumed as the point of nee Itshould be remembered that assumption of the point oforigin at any pia not influence the trend values. However, the values of'a’ or ‘b’ or both may" (2) XX is obtained by adding deviations (X), (3) Deviations are squared up (X*) and their total (EX°) is obtained. —_ .ANALYSIS OF TIME SERIES a deviations (X) are multiplied with the respective values (Y) and the aod out ( values (Y) are added to get ZY, ge calculations the values of ‘a’ and ‘b’ are obtained by using is”. inal i) equations and by solving them by the process of simultaneous 1 the eee é Y= Na + bX li) EXY = aEX+ bEx? si) yalues of and are placed in the trend equation (Y, =a + 6X) and the ()\* are found for the various values of X's, f yalues 1s roto? 1 or trend values by least squares method from the following time series data 9006 88 origin : “S 2007 2008 2009 2010 2011 2012 soles (Lakh Tons) 5 7 9 10 12 17 Trend Values ¥,=244+2.17X 5 2442.17%1=4.57 14 244217 27 244+217x 40 2.44+2.17x 60. 2.4+217% 2.4 +217 x 6 = 15.42 60=6a+21b (By Formula: EY =Na + bX) 248 =21a + 91b (By Formula : EXY=a2X + bEX?) Multiplying equation (i) by 7 and (ii) by 2 we get, 420 = 42a + 1476 496 = 42a + 1826 _76=-355 or 36b=76 nb=Ba2i7 Substituting the value of b in equation (i) 60 = 6a + 212.17 or 60 = 6a + 45.57 14.43 or 6a= 14.48 .a=— = 24 6a = 60 — 45.57 ¥,=24+2.17% : Mustration 13 oat sesh ine trend tothe fillowing dat by tomato fact epits ahs 8 origin, and (ii) 2010 as origin. Estimate the,trend value for 2013 : Ae : 2008 2009 2010 2011 2012 G@akhe) =: 45 56 78 46 15See SS asivess srarisics ___——_———. Solution i Determination of Trend Values By Least Squares Method (Origin 2007 = 0) 45+5x1=60 © 45+5x2=55 45+5x3=60 45+5x4=65 45+5x5=70 2010 2011 2012 N=5 300=5a+165 (By Formula : ZY =Na + bxX) au) 950 = 15a +556 (By Formula : EXY = oBX+ bEX") eT ee) Multiplying equation (i) by 3 ‘and subtracting it from equation (ii) 950 = 15a + 55b 900 = 15a + 456 50= 105 2 b=5 Substituting the value of 6 in equation (i) 300 =5a+15x5 —5a=- 300+ 75 5a =225 -.a=45 Y,=45+5X (year 2007 as origin) of Trend Values By Least Squares Method Determination (Origin 2010 = 0) Year x x xg xY Y,=a+bX 2008 45 =2 4 -90 60+5x-2=50 2009 56 a 1 -56 60+5x-1=55 2010 78 0 0 0 60+5x0=60 2011 46 1 1 46 60+5x1=65 2012 15 2 4 150 60+5x2=70 wes | sy=300 | =X=0 | sx2-10 | =X¥=50 a= 2 = 90 - 60,0- a= 7975 ¥, =60+5X (year 2010 as origin) 2 Caleulation of Trend Value for 2018 ; (a) Taking 2007 as origin : Y, +5X Yoog =45 + 5x 6 =% 75 Lakh (b) Taking 2010 as origin : Y,=60+ 5X Yoo1g = 60 + 5x 3=% 75 LakhANALYSIS OF TIME SERIES 595 2001 2004 2006 2007 2009 2010 2012 14 thod pis end values by met of least squares from the data given below : o 4 : 67 cS 15 68 65 60-64 41 ra Calculation of Trend Values Y,=a+bX 36 60-3x-6=78 9 60-3x-3=69 ; 68 -1 1 - 68 60-3x-1=63 a 65 0 0 0 60-3x 0=60 009 80 +2 4 100 60-3x 2=64 es AO 9 162 60-3x 3=51 41 +5 25 205: 60-8x_5=45 EY, = 420 BY _420_ py , 2A _— 252 N= Ges we elie as Y, =60 3X Note: There are two specific points in this Illustration : (1) If years are not continuous, this fact will have to keep in mind, while taking time deviations (X). (2) The value of'b'is negative. Tt means that there is trend of decline rate in place of growth rate. Tlustration 15 Fita straight line trend to the following data by the method of least squares after summing the quarterly data to yearly data and estimate the value for 2002 : S Quarter a | @ @ Qa 2010 15 2, = a 2011 18 30 me 2012 27 2, = a 2018 18 23 = 2 2014 M7 38. = a A i 2 18 SS Es 35 18 Silution The data in annual for quarters as follows : Year Operations : 1993 15 +26+28 +11=80 8; 1994 18 +30+32+10=90 90 1995 27+25+26+14=92 22ed 596 BUSINESS STATISTICS 1996 18 + 22+ 31+12=83 83 1997 174+ 284+ 34+15=94 94 1998 23 + 31+ 27+ 18=99 1999 21+ 28+ 30+ 13 = 92 Computation Table (Short Cut Method) Trend Values atbx a 90+2x(—3)=84 joes 90+2x (-2)=86 1995 90 +2.x (- 1)=88 eee 90+2x0=90 1997 90+2x1=92 1998 90+2x2=94 1999 90+2x3=96 630 Normal Equations : ysathx Zy _ 630 _ Nepales ty=Na>a= 56 af _ Say _ 56 _ and Exy = bEx? = b= > =9g=2 :. Trend line is y,orT=a+bx=90 +2n Substituting the values of u, we get the required trend values. Estimate for 2019 Yon1q = 90 + 2(2019 - 2013) =90+2x6=102 That is the value for 2019 is 102 thousand. The number of terms is even ; When the number of terms is even, take deviations about the mid-point between the years. To ease the calculations, multiply these deviations by:2 (or divide by 0.5). Iustration 16 ; Ts sales ofa company are given below : Estimate sales figures forthe yeas 2027088 an equation of the form Y = ab*, where X deviation from suitable year, Y=sales: Year : 2010 “9011 2012 +2018 «2014 2015, 2016 Sales (Lakh 2: 32 47 65 92 132 190 275~45153 9 ~ 3.3442 4 ' ~18129 o's 0 ; ao i te 2.1206 1 - 4.5576 4 0 7.3179 9 y=Nloga => logq=2l0RY _ 13.7926 log : a: at = L704 =logbEX’ =>log b= log¥ _ 43237 sx log ¥=log og b be ag = 0.1544 Betimate of sales for the year 1993 : Y=a* x log ¥=loga+X logb a log Y= 1.9704 +4x 0.1544 {X= 1993 - 1989 =4} = 2.5880 . Y= antilog (2.5880) = 3873 That Conversion of Annual Trend Equations into Monthly Trend Equations Tfannual trend equation is known, it can be converted into monthly trend eatin by dividing the computed constant'‘a’ by 12 and the viaue of B by 144. | ‘tis divided by 12 because it is the sum of 12 months (one year), while ‘b’ is divided by 144 because firstly itis converted into monthly viaue (dividing by 12) ad then again by 12 to make it monthly increment viaue. Thus the annual trend equation Y = a + bX is converted into monthly trend equation as follows : a,b Y=35+ qa Musiration V7 : i. Convert the following annual trend equation on a monthly basis : | Y=30+36% ., tiny se convert an annual trend equation to monthly basis, the vlaue of ‘a is by 12, while the viaue of ’ by 144. Thus, 30, 38 y Y=i9*i4 ® Y=2.5+ 0.025% Fitting a Parabolic or Non-linear Trend Perce Gals, in | vagttte may be many such conditions t the long term tendency in its a ee ere tie ere astnccrre ct Oe& aes BUSINESS STATISTICS powers (second, third or fourth, i i de; 0 , ete.). In this context the equation of second gree parabola or parabolic curve of the second degree is a8 follows : bie Y=a+bX+cX 7 Bjorke ‘a’ is the Y intercept, ‘p’ is the slope of the curve at the origin and ‘c’ ; ae rate of change in the slope. The value of a, 6 andc can be determined by solving the following three equations : EY =Na + bEX + cZX" EXY = a5X + bEX" + SxtY = aX? + DEX? + 02K" .Aiii) If time origin is taken from the middle point of the series, the value of ZX would be zero and the above equations are reduced to the following : EY=Na+cEX" EXY = bEX” $x°Y = aX" + cEX* Note : When EX = 0, EX° will also be zero. Illustration 18 Fit a second degree parabola for the following data : Year 5 2008 2009 2010 2011 2012 Value By 10 12 13 10 8 Solution cond Degree Parabola yx2210 | 2x®=0 | =xt=34 and £X° are equal to zero, 50 simplified N=5 | sy=53 In the above calculations 2X equations will be used : sY=Na+cEX* or 63 = 5a + 10e ld) EXY = bEX" or -6= 106 audi) =x?y =cEX?+cX* or 94= 10a + 34c Gi) As per equation ii): Sees a 10b=-6 «b= Gq =~ 0.6 Multiplying equation (i) by 2 and subtracting it from equation (iii) 94 =10a + 34c 106 =10a + 20c es -12=14¢ or cay,ANALYSIS OF TIME SERIES tating the value ofc in equation (i) 7 gab 53 = 6a + 10 x- 0.857 53+ 8.57 = 5a or 6a =61.57 a= $1.57 = 12.314 Y=atbX+cxX" = 12.314 + (- 0.6)X + (- 0.857)X2 Y = 12.814 — 0.6X - 0.857X" Calculation ee = 12.314 - 0.6 x- 2- 0.857 x- 2" = 10.086 = 12.314 -0.6x-1-0.857x- 1? = 12.057 9010 =12.314 =12,314 9011 = 12,314 - 0.6 x 1-0.857 x1? = 10.857 2012 = 12.314 -0.6x2-0.857 x2 = 7.686 Merits of Least Squares Method 1. Completely objective: This method is completely objective, in which trend values are calculated on the basis of well-defined mathematical principles and formulae. There is no possibility of personal bias in this method. 2, Forecasting * The equation of a straight line based on least squares establishes 2 functional relationship in between x and y series and through this relationship forecasting can easily be made for future values. 3. Trend for the entire period : In moving average method trend values cannot be found out for the entire period while the method of least squares gives the trend values for the entire time period. 4, Line of best fit : The trend line obtained by this method is the line of best fit, because the sum of positive and negative deviations of original data from this line is zero and sum of squares of deviations is minimum. 5. Calculation of change rate : If data are on yearly basis then annual growth rate or decline rate can be obtained by this method. Limitations of Least Squares Method ie 1 Tedious and complicated : This metho Point of mathematical calculations. it 2. Lack of flexibility : even a single value is added or deleted to the series, ee necessary to do all the computations again and generally the tion of trend also does change. e 3. Wrong selection of equation : If the selection of trend equation (linear, or some other type) is not proper, it may lead to fallacious results. 4. Limitations of predictions : Prediction in this method is based on ao trend and the impact of seasonal, cyélical or irregular variations is dis tedious and complicated fromANALYSIS OF TIMES SERIES es trend cannot be curvi-linear, f ie moving average values for different years are called trend. & that a-22 _i? tosay andb=— » i correct sat Say ia gonual trend ey cannot be converted into: monthly or quarterly values. ® nu possible to interpolate or extrapolate values with the help of a straight line ft WH 7008. cue, 2. Folse, 3. False, 4. True, 6. per aoe areal » 5. Ture, 6, False, 7, False, 8, True, 9. False, 10. 601 tin ee cae ost important factors causing seasonal variations are : = short-term variations are classified as ; a qhe Jong-term tendency of increase or decrease in time series is known as ......--. 1, Thetime series data exhibits.......... trend if the rate of growth is constant. Phe least squares trend equation y = a + by exhibits ........ trend, ifb > 0 anda....... 5 trend, if b <0. F Jo multiplicative model of time series y =....... , Seasonal variations are the short-term variation with period... g. Theregular movements occurring in a period of more than one year are called as _.. fluctuations. g, Thetotal of squares of differences between original values and trend values by the method of least squares is........ , 40, O-T-S=--— 1. ‘Atime series consists of data arranged........ : 1. The additive model of a time series is expressed as, 18, The line obtained by the method of least suqares is known as the line of ..... . 14. Repetitive and predictable movement around the trend line in 1 year or. lessis.....- variation. ieee variation is the time series component that oscillates above and below the trend line or period longer than one year. [Ans.1. Weather or season and social customs; 2. (a) Seasonal, (b) Cyclical; 8. Secular Trend; 4. Linear; 5. Increasing, Decreasing; 6. Y=7xSxCxJ; 7. Less than one year; 8, Cyclical; 9, Minimum; 10. C +I, 11. Chronologically, 12.7'+ S +C +1, 18. Best fit, 14. Seasonal, 15. Cyclical] PRACTICAL QUESTIONS Free-hand CurveMethod 1, Draw a trend line by free-hand curve method from the following values : Year : 2007 2008 2009 2010 2011 2012 Value : 20 24 22 30 28 32 Semi-AverageMethod 2 From the following data fit the tsight ine rend by the method of semi-averages: fear 2008 2009 2010 2011 2012 a, Profit before Tax in Lakhs) 228) 30. 31 29 31 Fit a trend by the method of semi-averages from the information given below : Year : 2004 2005 2006 2007 2008 2009 2010 2011 2012 Sales(@Thousands) : 18 24 26 28 33 36 40 44 48——HUSINESS STATISTICS 602 ada AverageMethod is i Moving. vis made by a table fa manufacturing concern in Eastern Uttar Pradesh 4 Be cop to 2012 are given 28 below (in lakhs of rupees) : SS ee uring a Export Year Export eta 361 2007 420 an 366 2008 450 ens 361 2009 500 2003 362 2010 518 2004 400 2011 540 419 2012 557 Ks z aes ta on a graph paper and determine the trend by using five-year oe aa 368, 381.6, 390.4, 402.2, 419.8, 439.8, 459.6, 485.6, Rae Plot the above di moving average. : 5. Calculate the 4 yearly moving average of the given data : ty ia ee Index No. _Year Index No, sea 52.7 8 87.2 2 19.4 9 79.3 3 76.3 10 103.2 4 5 6 [Ans. —, —, 70.688, 74.375, 79.725, 85.925, 89.918, 92.475, — —] 6. Using three-yearly moving averages, determine the trend from the following data— Year :1 2 3 4 5 6 7 8 9 10 11 12 13 14 Production: 11 12 9 8 14 15 18 16 17 19 21 24 28 30 [Ans. —, 10.67, 9.67, 10.33, 12.83, 15.67, 16.33, 17, 17.38, 19, 21.83, 24.88, 27.83, 7. Determine the trend values by three yearly moving averages : Year : 2007 2008 2009 2010 2011 2012 2018 2014 2015 2016 Variable : 8 12 10 138 15 12 146 WWM [Ans. —, 10.00, 11.67, 12.67, 13.33, 14.33, 15.00, 15.67, 16.00, —] 8. Calculate the three yearly moving averages for the following time series and plot them with the original figures on the same graph to show its secular trend : Year ak 2 3 4 5 6 7 8 9 10 Annual Output: 110 104 98 105 110 120 115 110 114 125 [Ans. —, 104, 102.3, 104.3, 111.7, 115.0, 115.0, 113.0, 116.3,—) 9. Find out the value of the trend by 3-yearly moving average and plot them ona graph paper : & Year : 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Value: 10 15 12918 15 22 19 24 20 26 22 30 2% (Ans. —, 12.3, 15, 15, 18.3, 18.7, 21.7, 21, 23.3, 22.7, 26, 25.7,—] 10. Find out the value of 4-yearly moving averages and plot them on a graph paper. Draw the trend line with help of these trend values ; ‘Year 2005 2006 2007 2008 2009 2010 Production (Lakh ton) 5 6 7 7 6 8 Year 2011 2012 201 2016 Production (Lakh ton) 9 10 oot OH (Ans. —, —, 6.38, 6.75, 7.25, 7.88, 8.62, 9.25, 9.75, 10.12, —,ANALYSIS OF TIMES SERIES 603 : information about tre1 tT : ov e a Rae nd using four-yearly moving averages and repre- ses Poags 2006 2077 2008 20g0 20 90 gz? HEP 2BN* Zaye Y * “{ans.— ae 16.76, 17.25, 17.88, 18,62, 19.25, 19.75, 20.12,—, —] - oa four-yearly cycle, calculate the trend by i r sf yy the method of movin; ay rs Deen lowing data relating to the production of tea in india: ibd Year : 2007 2008 2009 2010 2011 2012 2018 2014 2014 2014 n@akh kg): 464 515 618 467 602 540 557 571 586 612 (Ans. —, —, 495.8, 503.6, 511.6, 529.5, 553.0, 572.5 — —] Least Squares i otbod of sta straight line trend by the method of least squares to the followini data.Also gt tps orginal data and trend values ona Se ee Yeor j : 2007 2008 2009 2010 2011 2012 ” Sets (in thousands) saa 10 12 14 17 24 soles of T- V- tans. ¥e= 14+ 1.54X (2009.5 = 0), Trend Values = 6.3, 9.98, 12.46, 15.54, 18.62, 21.701 1“ Find trend line to the following data by the least squares method : Yeor > 2008 2009 2010 2011 2012 ic a 18 21 23 27 16 Estimate the production in 2013. [Ans. ¥.=21+0.2X (2010 = 0), Trend Values =— 20.6, 20.8, 21, 21.2, 21.4; Estimate of 2013 = 21.6) 15. With the help of the following data fit a straight line trend by the method of least squares and estimate the trend of sales for the next two years : Year ; 2007 2008 2009 2010 2011 2012 Sales (% Lakhs) : 120 140 150 170 - 190 200 [Ans. Trend Values = 120.97, 137.25, 153.53, 169.81, 186.09, 202.37; Trend for next two years = 218.65, 234.98 7 Lakhs] 16. Below are given the figures of, production (in thousand tons) of a sugar factory : Year : 2006 2007 2008 2009 2010 2011 2012 Production hea 88 94 85 91 98: 90 (a) Fit straight line by the method of Teast squares’ and show the trend values : (b) What is the monthly increase in production ? [Ams. (a) ¥e=89 + 2X (2009 = 0), Trend Velues = 83, 85, 87, 89, 91, 93, 95; (b) Monthly Increase Rate = 2000 - 16667 Tonsl ast squares method and also Tl. Find out trend values from the following data by Je find the estimated value of 2017 : 5 apa 73006 2007-2008 «2009-2010 2011 2012 Sales (Tons) eee Pee ee 858 [Ans :¥-=76 +4.86X (2009 = 0), Trend Value = 61.42, 66.28, 71.14, 76, 80.86, mf 85.72, 90.58; Estimate for 2017 = 114.88] Fit a straight line trend by the method of least squares to the following data. Ye the profit for the year 2014: oo + 2005 2006 2007 2008 2009 2010 2011 2012 Profit(@ Lakhs) : 38 40 65 72 69 60 87 95 TAns : ¥.= 65.75 + 3.67X (2008.5 =0), Trend Value = 40.06, 47.40, 54.74, 62.08, 69.42, 7676, 84.10, 91:44; Estimate for the year 2014 = 106.12 % Lakhs]19, 21, LK BUSINESS STATISTICS Below are given the figures of sales (in thousand 8) of a concern. Calculate trend by the method of least equares. Show the trend on a graph paper." Year : 2005 2006 2007 2008 2009 2010 2011 2012 Sales : 60 64 70 80 86 9 100 a [Ans : 61.03, 66.45, 71.87, 77.29, 82.71, 88.18, 03.55, 9g i , fit a linear trend From the following figures of output of a sugar factory, fit nd by leagt squares and show ihe trend line on a graph paper. What is the monthly increase in production ? me 2006 2007 2008 2009 2010 2011 2012 Production (000 Qatls.) 2 10 iW 11 618) (5 [Ans : ¥- = 13 + 0.75X (2009 = 0), Trend Values = 10.75, 11.30, 1225,15, 13.75, 14.50, 15.25; Monthly Increase Rate =" = 62.5 Qug) ‘aking icin, find out a straight line trend equation Taking tae yee varea, Eotimate the trend value for the year 2004 ; by the Year : 2003 2005 2006 2007 2008 2009 2012 Value : 140 144 160 152 168 176 180 [Ans: ¥-= 159.29 + 4.955X, Value for 2004 = 144.495) ‘The following table shows the production of room-coolers by a group of manufac. turers from 2006 to 2012 in lakhs of rupees : Year : 2006 2007 2008 2009 2010 2011 2012 Production i 120 195 216 207 180 261 285 t the above data on a graph paper. : ‘ % a a straight line trend to the above data using the equation Y=a+bX and show the trend line on the same graph paper ? (©) Estimate the production of: room-coolers for 2013. [Ans : ¥. =209.14 + 21.11% (2009 = 0), Trend = 146, 167, 188, 209, 230, 251, 273 Production of coolers will be in whole numbers Estimate for 2013 = 294 Coolers) Using least square method, calculate the following with the help of the data given below : (i) Trend values (ii) Monthly increase in production of iron (iii) forecast the production of iron in the year 2014, 2015 and 2016. etal a a fan 2s a i rs : 700 peice oer and @) 144.77, 286.67, 357.62, 428.57, 499.52, 570.47, 712.37 Gi) 5,912.5 tons per month (iii) 854.27, 925.22, 996.17] Fit a least square trend to the following data. Also estimate the volume of sale for the year 2015; Year : 1985 1990 1995 2000 2005 2010 Sales (€ Lakh) = 12 15 17 22 24 30. {Ans. Trend = 11.29, 14.77, 18.26, 21.74, 25.23, 28.71; Estimate for 2015 = 32.201 Lakh ®t] Using the equation of ¥=a + bX, fit a straight line trend for the data given below relating to production (Lakh %) and show the trend line on a graph paper: Year 2006 2007 2008 2009 2010 2011 2012 127 102 130 152 126 142 138 [Ans.¥, = 131+ 3.89X (2009 = 0)) From the data given below find secular trend by least squares method and estimate the sale for 2014 and 2016 : Year > 2008 2009 2010 2011 2012 Sales ( Lakh) : 100 120 110 140 180 (Ams. ¥. = 130 + 18x (2010 = 0), Trend Value = 94, 112, 190, 148, 166; ¥oo14= 202, Yoos= 238)ANALYSIS OF TIMES SERIES _yen trend equation ¥,=90+25X, origin : ing wend ae ce 2004-2012 1 ae gees ieceee eee Ss = we tion ofa fetil #6, 86.25, 87,5, 8875, 90, 91.28, 9250, 93.75, 95 Bie Sone factory (in thousand tons) is as follows : : 10 15 2008 2009 «2010 ©2011 +2012 90 98 85 91 100 raya straight line trend by the method of least squares and tabulate the trend @ ues ; val ert your annual trend equations into a monthly trend equation. Con’ @ a UAns. Ye = 87+ 42X (2008 = 0), Monthly Trend Equation) assed degree parabola curve method in the following data and make forecast # Reoext three yoRNe oy a : 2 wes 2009 2010 2011 2012 sors cae 7000) 7 4 8 6 [Ans, Taking 2010 as origin ¥,=6+03X) ria second deeree parabola to the following data and calculate the trend values ® fprtbe gifferent years. Also give your estimation for the year 2014 and 2015: mae 3 2008 2007 2008 2009 2010 2011 2012 Bis 100 164-196. «Ss 212s «220s 22226 [Ans. Trend Values 107.93, 155.99, 191.97, 215.87, 227,69, 227.43, 215.09; 2014 = 154.17, 2016 = 105.59) so price of a commodity during 2006-2012 are given below. Fit a second degree estimate the price of commodity for the year 2014. 2012 2 parabola and Year : 2006 2007 2008 2009 2010 2011 Price ; 100 107 128 140 181 242 362 [Ans : (Origin 2009) a = 140.8, 6 = 39.61, c = 9.8; Price of 2014 = € 583.85) $2. Thesales of a company in lakhs of rupees forthe year 2008 to 2012 are given below. Estimate Ges for the year 2013 using an equation ofthe form Y= ab": Yerr(X) : ~—«-2008 2009 2010 2011 2012 Sales (Y) 65 92 132 190 275 Lakh &, log a = 2.1231, log b= 0.1568] [Ans. Estimate for the year 2018 = 392.19