Page 120 - Contributed Paper Session (CPS) - Volume 7

CPS2043 Adnan Dawood K. B. et al.
For classifying the households whether they belong to HWL or to the LWL,
we can build the first linear regression formula based on the suggested
independent variables depending on the whole population, in order to predict
whether the household belongs to HWL or LWL. Households with predicted
annual per capita expenditure above the CP_1 are non‐poor and classified as
HWL. No more work is required with this category of households because they
are not poor.
Based on the households belonging to the LWL and a trend line on the
natural logarithm curve a second cutoff point (CP_2) should be selected
carefully, and then the population is divided into two segments; the Middle
Welfare Level (MWL) and the Rest of LWL. Next, we build the second linear
regression formula based on the suggested independent variables for the LWL
households. This formula is to predict whether the household belongs to the
MWL or to the Rest of LWL. Households for whom the predicted annual per
capita expenditure is above the CP_2 are classified with MWL. These
households are lying between CP_1 and CP_2. A third regression formula
should be built for these households which lies in the MWL, this formula is for
predicting the annual per capita expenditure for these households belonging
to MWL.
In order to divide the households belonging to the Rest LWL into two final
segments; Lower Middle Welfare Level (LMWL) and the Bottom Welfare Level
(BWL), a third cutoff point (CP_3) should be selected based on the abject
poverty line and the trend line on the natural logarithm curve. The fourth
PMTF should be build based on the households belonging to the Rest of LWL.
Households with a predicted annual per capita expenditure above the CP_3
are classified as LMWL and the other is classified to the BWL.
A fifth specific PMTF should be built for LMWL households, based on those
households lying between CP_2 and CP_3 to predict the annual per capita
expenditure. Lastly, in order to predict the average annual per capita
expenditure for the households lying in the BWL (below CP_3) the sixth linear
regression formula through the origin; based on the suggested independent
variables for the BWL, is built.
At the end of construction process for the formulas of the PMTF, we got
six formulas: three formulas for classifying the households between the four
categories: high welfare category, middle welfare category, lower welfare
category, and the bottom welfare category. And we have another three
formulas for predicting the annual per capita expenditure consumption; this
predicted expenditure will be used as a final prediction for the household to
decide whether it is poor or not.

107 | I S I W S C 2 0 1 9