Facing the opening of electric power market

customer demand has been highly valued in order to enhance the competitiveness of power grid enterprises. Power outage is the primary factor affecting customer perception. The identification of customer outage sensitivity is one of the important conditions for accurate customer service. Improving the identification accuracy of customer outage sensitivity is the key to improving customer service level.A method to improve the accuracy of outage sensitivity model is provided. Based on multi dimensional indicators such as customer information and behavior

and on the basis of traditional entropy method

adaline algorithm is used to modify the weights

and unsupervised learning is transformed into semi supervised learning

so as to enhance the scientificity of weight assignment.