dc.description.abstract | The problem of Time-series is of evaluating the non-randomness present in the series in a manner that pre supposes nothing as to the nature of non-randomness.
Significant work has been done with this objective
in view by Kolmogorov [10], Bartiett 13] ) Heiner 115], Durbin [8] ,Tutkey [12] , 113] , Box and Jenkins [6] ,Parzen [11] , Watson and Hannan [9] and very recently by Bhansali I4]
Still the problem remains as exciting as ever.
In this dissertation, emphasis has been laid on examining the physical aspects of non-randomness present in the series, leaving aside the problem of prediction. This has been done in Chapter V by taking an actual data on Kenya's export figures and trying to investigate the cycles that occur in the export trade. Two methods of eliminating tren~ have been worked out and the remaining series exemined by using the power
spectrum analysis.
Chapter I introduces the concepts of time-series. Chapter II deals with various filters that are used in time- series analysis to make the series stationary. Chapter III
illustrates the time-series as an auto-regressive and a moving average
stochastic process. Chapter IV outlines the tests of significance.
The objective throughout has neen to use the available
1iterature and where necessary to develop the existing material to investigate the salient features of Kenya's export. | en |