Estimating True Movement In Time Series As Linear Combination Of Two Indicators

Abstract

This paper describes an iterative procedure for estimating true change in a time series as a linear combination of two imperfect indicators with weight chosen to minimize errors particularly when the time series (dependent variables) is not or cannot be observed, this research work was compiled to estimate the coefficients of the ?’s responsiveness of indicators to change in true series, in order to determine the equation that connect variables. It also describes how an appropriate weight could be chosen for the two indicators as to error.

We found that; the best estimation for changing the number of caesarean section is;

T₁ = 0.1720x + 0.967y₁- 0.0287

The length of stay in hospital series received a substantial weight compared to the weight of parity series. The final estimate of a_x to be positive and reduction of error expected from using the proposed procedure was estimated at about 6% and 27%  per month compared to both parity and length of stay in hospital indicators respectively.

Key Words: Imperfect Indicators, True Series, Weighted Time series