Abstract

Wheat is considered as an important production in agriculture. Because it is included among the basic necessitates for the human livings. Therefore, the production of wheat is important in various field. Especially, in Pakistan there is a need to increase the production of wheat because the population is increasing. There are various factors effecting on the wheat production. The need is to focus on the factors related to wheat. The current study focused on the extraction of factors related to the wheat production. Dataset is used from Statistical Bureau of Pakistan consisting of 466 observations in the analysis. In which 80% dataset is taken as the train dataset while the 20% is kept as a test dataset. 46 predictors related to the yield of wheat is observed with the dependent variable as yield of wheat. Econometric issues such as multicollinearity and outliers are observed in the dataset. For this purpose, hybrid model of robust estimators and forward stepwise regression is used. Among the robust estimators, huber M, hampel M and bisquare M is chosen for the comparison purpose. The results showed that the hybrid model of forward stepwise and hampel M estimators provide the efficient results in term of minimum mean square error (MSE) and mean absolute percentage error (MAPE).