考虑评论感知情感和宏观经济指标的汽车销量预测

为提高汽车销量的预测精度,从而为企业估算销售收入、制定生产计划和进行发展战略决策提供有效支撑,提出了一种考虑评论感知情感和宏观经济指标的汽车销量预测方法.首先通过构建领域专有情感词典,对口碑评论进行情感分数提取,其后基于前景理论和归因理论对评论情感的影响机理进行研究,获得评论感知情感指数,最后结合经济指标和历史销量,通过线性回归模型对真实汽车车型销量进行预测.结果表明:与仅使用历史销量数据相比,模型中加入评论感知情感和宏观经济指标有效提升了销量预测精度.

Forecast of Automobile Sales Considering the Perceived Emotion of Online Reviews and the Macroeconomic Indicators

In order to provide effective support for enterprises to estimate sales revenue, formulate production plans and make development strategy decisions, a method considering the perception of comments and macroeconomic indicators is proposed to improve the accuracy of automobile sales forecast. In this method, a domain-specific dictionary was firstly constructed to quantify the sentiment intensity of reviews. Subsequently, the sentiment intensity was adjusted to the index of perceive emotion of reviews with the prospect theory and the attribution theory applied. Finally, combined with national macroeconomic indicators and historical sales, a linear regression model was built to forecast the real automobile sales. The result shows, compared with only using historical sales, adding the index of perceive emotion of reviews and the macroeconomic indicators to model can effectively improve the accuracy of forecast.