Electricity demand forecasting is an essential process for electricity planning, designing strategies and recommending future energy policies. The changing behavior of the socio-economic growth beside the incomplete coverage of the environmental impacts can make a long-term energy demand forecasting process for a specific energy network challenging. This article presents four new developed multiple regression models for Electric Energy Peak Load and the main affecting factors for Kingdom of Bahrain as a case study. Time series analysis of seven years monthly load data was conducted. The method was hybridized with Machine-learning tools to find suitable forecasting linear and non-linear models for Bahrain electricity network. Residual analysis was adopted to find the model that best fit the Peak load data. Cross validation aims to evaluate the efficiency of a predictive model. For this purpose, a new peak load data set for an eight year was gathered and tested. Results are reported to guide Bahrain electricity network forecasting needs for the next future years. The developed technique can be extended to the hybrid renewable energy system that Bahrain and other countries in the region has recently announced to adopt. © 2018 IEEE.

View