基于GBDT算法的电视剧收视率预测

Audience Rating Prediction Based on GBDT Algorithm

准确的收视率预测具有较高的商业价值,能够降低各方投资风险同时提高多方收益,形成合作共赢。为此,基于梯度提升决策树(Gradient Boosting Decision Tree,GBDT)算法建立电视剧收视率预测模型。研究表明,基于影响因素划分的GBDT电视剧收视率预测模型能够有效预测不同主创团队、题材及热度的电视剧的收视率,模型预测值能有效拟合真实收视率。GBDT预测方法为电视剧播前收视率前预测提供了一种新的思路。

Accurate forecast of audience rating has high commercial value, which can reduce the investment risk of all parties and improve the income of all parties, thus forming win-win cooperation. Based on Gradient Boosting Decision Tree(GBDT) algorithm, TV series ratings prediction model is established. The research shows that GBDT TV ratings prediction model based on influencing factors can effectively predict TV ratings of different creative teams, themes and heat, the model predicted value can effectively fit the real audience rating. GBDT forecast method provides a new way of thinking for TV series ratings forecast before broadcast.