Regression Model Of Household Consumption Expenditure

Regression model of Household Consumption Expenditure.

In the table below (6th Table) you can see significant variables and one outlier.

To create a model, I need to use a backward method of elimination. Using this method I will start working with all variables and then delete some of them, depending on its p-value.

In the table above (8th Table) one of the sectors should be removed and the regression analysis should be repeated without it. This sector is called „ Main source of income of household head: Employed in public sector, its valu eis equal to 0.45 and is bigger than 0.05, so that is the reason why it had to be deleted.

The table above (9th Table) shows that another sector has to be deleted because it has a p-value equal to 0.27 and it is bigger than 0.05. This sector is Socio-economic group of household head: Retire people.

The table above (12th Table) shows us that our R2 value is equal to 0.66 and it is higher than 0.6. so it means that we got a good result.

First situation: x1=1, x2=1, x3=0, x4=0, x5=0 (Household head has primary education, is married and there are no outliers).

Second situation: x1=1, x2=0, x3=x4=x5=0 (Household head has primary education, is not married and there are no outliers).

The MAPE is equal to 35.96%. My forecast is off by 35.96%. Less than 5% of MAPE is a good result, but I didn‘t get it and it means that my model is not good.

Errors are random and their average is almost equal to 0 (Chart 2).

Six independent qualitative variables and three outliers were used in the regression analysis

MAPE was equal to 35.96%, the forecast was off by 35.96% and it means that it did not work.