Abstract:This paper is based on the marketing system data to conduct electricity forecasting, analyze the trends of electricity consumption in different regions, industries and periods, drill down into the data, find hidden information and use it. The factors affecting electricity consumption are excavated in all directions, and the relevant factor formation factor reserve library is selected to prepare for subsequent modeling. Through the use of time series, multiple linear regression and gray prediction algorithms to predict electricity, enrich power forecasting methods, improve short-term, medium-term and long-term power forecasting capabilities, and visualize the forecast results to provide reliable data support for power forecasting. The example proves that the method can effectively improve the accuracy of power forecasting and achieve accurate prediction of power consumption in the future,which not only improves customer service capability, improves the work efficiency of customer service departments, but also effectively improves the core competition of power companies, So that power companies can make more wise business decisions in the increasingly fierce market situation and seize the leading position in the electricity market.