Construction of Potential Customer Index System Under Machine Learning Algorithm

Compared to traditional data, data consisting of non-traditional information or data acquired through non-traditional means is big data with new sources, dimensions and breadth. In order to evaluate micro and small enterprises in multiple dimensions, potential customer evaluation indicators are constructed by integrating traditional and non-traditional data to ensure the comprehensiveness and diversity of data dimensions, which is conducive to the service platform to explore potential customers. Taking the actual business data of a service platform as the research sample, based on the two-stage feature selection of AMI-CMI-RFE, common classification algorithms such as random forest and decision tree are selected, and the effectiveness of the two-stage feature selection is tested in terms of the four aspects of accuracy, precision, recall and F1 value. The experimental results show that relative to the original data, the four performance indicators of the two-stage feature selection algorithm with AMI-CMI-RFE are significantly improved, indicating that the algorithm is effective for potential customer mining.