Apriori-based prediction of the multi seat elections

摘要

Election forecasting has attracted many research efforts. Election forecasting helps the government to optimize the rules of election and candidates to adjust their campaign strategy according to forecasting results, and it contributes to election campaigns. Based on the poll model, this paper proposes poll model improved by delegating and weighting operations applicable to the election of Hongkong legislative council. Based on Apriori algorithm, the model predicts election results of the Hongkong Legislative Council in 2016 and makes a comparative analysis of the actual election results of the legislative council in 2016. From the analysis of these two models' forecasting results regarding the 2016 Legislative Council election result, we can see that an improved poll model by delegating and weighting operations can accurately predict the local elections in Hongkong. The forecasting accuracy of the improved poll model by delegating and weighting operations is 82.5%, which is 10.9 percentage points higher than the poll model. The improved poll model by delegating and weighting operations can predict election results of Hongkong Legislative Council more accurately, which fills the gaps in the study of election forecasting studies on the Hongkong Legislative Council elections and has positive significance for the long-term and stable development of Hongkong.