This study investigates the relationship between the production, exportation and consumption of electricity in Nigeria from the period of 2000 to 2011. This shows how the knowledge of analyzing residuals can help in developing a good model for prediction. The application was restricted to a linear regression model and it was developed for predicting consumption of electricity in Nigeria. Tests based on residuals analysis such as heteroscedasticity, multicolliniarity, and autocorrdation were applied to the original consumption-model. The model passed all the tests except the test for constancy of variance, thus suggesting that the disturbance terms are heteroscedastic which was later corrected by transforming the original data using reciprocal transformation and re-tested which eventually passed the test before it was considered adequate for prediction. At the final result, it was only consumption that was linearly related.
