The aim of this work is to analyze factors influencing electricity consumption in Japan using regression analysis. Every season regression models are developed for forecasting and determining elasticity coefficient associated with climatic conditions. As explanation variables, we use temperature, relative humidity and other factors such us holidays. Then, several statistical tests, for instance, t-test, adjusted coefficient R2, and Durbin-Watson statistic test, are used to validate the models. As an error analysis, MAPE is used to evaluate the forecasting performance. The results indicate that almost all explanation variables of the 4 seasonal models with different time period have significant influence. The models with autoregressive concerning the error term give better forecasting than without autoregressive term. Moreover, 2 scenarios are tested for each season to observe further the effect of changing climatic conditions to the electricity demand.