Electoral forecasting using a novel temporal attenuation model: Predicting the US presidential elections
作者：
Highlights：
• A new model for multi-opinion systems targeting better election forecasting.
• Temporal attenuation (TA) models opinion momentum from pre-election polls.
• Validation on US Presidential elections datasets from 1968 to 2016.
• TA achieves 23–37% higher forecasting performance than state of the art.
• Our method works without the need for social-economical indices.
摘要
•A new model for multi-opinion systems targeting better election forecasting.•Temporal attenuation (TA) models opinion momentum from pre-election polls.•Validation on US Presidential elections datasets from 1968 to 2016.•TA achieves 23–37% higher forecasting performance than state of the art.•Our method works without the need for social-economical indices.
论文关键词：election forecast,temporal attenuation,opinion polls,social media,US Presidential elections,computational intelligence
论文评审过程：Received 15 May 2020, Revised 31 March 2021, Accepted 23 May 2021, Available online 2 June 2021, Version of Record 8 June 2021.
论文官网地址：https://doi.org/10.1016/j.eswa.2021.115289
