Interruptions in the supply of energy to the entire system or part of it may be caused by a variety of weather or non-weather-related reasons. To minimize the inconvenience to and loss of productivity for hundreds of thousands of customers, electric companies must prepare for such possibilities and contingencies in advance so as to be ready with adequate resources, equipment, and manpower to tackle such interruptions. The cumulative losses to the business due to a single day’s temporary outage may add up to hundreds of millions of dollars as the work completely stops and also many products, such as food, get spoiled. This article takes upon the modeling and prediction of such interruptions using a large historical data. The post-shrinkage modeling provides a very articulate approach to more accurately predict the severity and extent of interruptions from various weather and non-weather-related data. We adopt a number of such approaches and suggest those which result in superior predictions. Through extensive simulations and resampling, we establish the validity of certain approaches.