Underfloor heating users prediction based on SVDD

Data analysis and utilization play an important role in the development of enterprises. How to extract valuable information from existing data is the focus of current research. Prediction of underfloor heating users is an important and urgent research topic of Gas Co. This paper constructs a prediction model to analyze whether the gas users are underfloor heating users or not based on the gas data sets. Because the training set we obtained only contains one class of data, we adopt the SVDD algorithm, which can effectively solve the one-class classification problem. In the experiment, we construct the prediction model effectively and estimate the proportion of underfloor heating users in gas users. Considering the sensitivity of the parameters in the SVDD algorithm to the prediction model, we obtained the relationship between the proportion of underfloor heating users and the values of parameters through the parameter tuning, which could provide the reference for Gas Co to select parameters.