Multiple linear regression models for prediction of rice production and productivity based on rainfall

A study on the influence of monthly total rainfall on rice production and productivity was carried out taking data for five decades from 1950 to 2003. Multiple linear regression technique was used to bring out a statistical model. The correlation studies revealed that there were no interrelationship between monthly rainfall and rice production parameters. In the absence of correlations, year variable was added in the study along with the monthly rainfall to take care of trend effects in rice production parameters. Step wise backward regression brought out that only year and July rainfall were contributing to rice productivity and production. The total rainfall of July for the years 2004 to 2007 were used to validate the models and proved for their accuracy.