解释决策林树评估模型 [英] Interpreting the Evaluation Model for Decision Forest Tree

问题描述

我正在研究一个项目，以提高生产力一个人在不久的将来，我正在Azure机器学习工作室中使用Decision Forest Tree.我有一个 大量的示例数据(5459行)和9个功能.我认为我的模型是过拟合的受害者，因为我对训练后的数据(即.64)实现了良好的确定系数，而0.34的确定系数几乎达到一半.另一个 我要关注的是MAE的价值，因为我正在努力使价格低于.10，但我认为这是不可能的.我要获得更低的MAE值和更好的COD的唯一方法是在复制样本数据时.我已经尝试过交叉验证 和Tune模型参数，但效果不佳.

I am working on a project to find Productivity of a person in near future and i am using Decision Forest Tree in Azure Machine Learning studio. I have a good amount of sample data (5459 rows) & 9 Features. I believe my model is victim of Overfitting because i am achieving a good Coefficient of determination for trained data i.e. .64 while the Coefficient of determination almost half for .34. The other concern is the value for MAE as i am trying to achieve the vale lower than .10 but i think it´s impossible. The only way i am achieve the lower value for MAE and better COD is when i am duplicating the sample data. I have already tried the cross validation and Tune model parameters but it´s not doing any good.

推荐答案

我们可以在AI画廊中引用很多实验，请参阅:https://gallery.azure.ai/browse/?skip = 0& orderby = trending％20desc& algorithms =％5B％ 22Two-Class％20Boosted％20Decision％20Tree％22％2C％22Boosted％20Decision％20Tree％20Regression％22％2C％22Multiclass％20Decision％20Forest％22％5D

We have a lot of experiments you can refer in out AI gallery, please see: https://gallery.azure.ai/browse/?skip=0&orderby=trending%20desc&algorithms=%5B%22Two-Class%20Boosted%20Decision%20Tree%22%2C%22Boosted%20Decision%20Tree%20Regression%22%2C%22Multiclass%20Decision%20Forest%22%5D

我也会调查您的问题，并在您遇到问题后与您分享.

I will also look into your problem and share with you once I have some.

此致

雨桐

这篇关于解释决策林树评估模型的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！