基于随机森林算法的鳄梨价格预测

Avocado Price Prediction Based on Random Forest Algorithm

摘要: 为了能够更好地预测鳄梨的价格走向趋势，解决在大量特征和大数据下价格预测精度低的问题。本研究在随机的基础上提出了一种基于Pearson系数的随机森林新的组合模型方法。首先，利用Pearson系数进行相关性检验，来进行特征筛选；对随机森林参数调优；最后利用剩余特征进行建模回归预测，并得出最终结论。实验结果表明：改进后的随机森林预测值的平均绝对误差(MAE)和均方误差(MSE)都得到了较大的提高。经研究发现，本文建立的新的组合模型，可以实现对鳄梨价格的短期预测，并且可以达到不错的预测效果。

Abstract: In order to be able to better predict prices to the trend of avocado, and solve the problem of low price prediction accuracy under a large number of features and big data, this study on the basis of random puts forward a random forest new combination model based on coefficient of Pearson method. Firstly, Pearson coefficient was used for correlation test to carry out feature screening; tuning random forest parameters; finally, residual features were used for modeling regression prediction, and the final conclusion was drawn. The experimental results show that the improved random forest predicted mean absolute error (MAE) and mean square error (MSE) got improved. The study found that through a new portfolio model, this paper can realize the avocado price short-term prediction, and can achieve good prediction effect.