基于机器学习算法的输电线路工程投资预测

技术方案深度的不足导致依据定额概预算来确定输电线路工程投资的方法准确性低、工作量大，因此，研究基于机器学习的投资预测模型需求迫切。针对输电线路投资的高维数、非线性等特点，提出了基于极端梯度提升(extreme gradient boosting, XGBoost)算法的输电线路工程投资预测方法。通过采用实际输电线路工程数据对模型进行训练和测试，预测结果显示XGBoost模型在预测精度、结果偏差方面相较于神经网络和支持向量机(support vector machine, SVM)都具有较大的优势，能输出指标重要性排序，为决策者提供有效的投资额和控制指标参考，且模型的可靠性和可解释性较高。

Transmission line project investment prediction based on machine learning algorithm

The lack of depth of technical solutions leads to the low accuracy and heavy workload of the method to determine the transmission line project investment based on the quota budget. Therefore, it is urgent to study the investment prediction model based on machine learning. According to the characteristics of high dimension and nonlinearity of transmission line investment, a transmission line project investment prediction model based on XGBoost algorithm is proposed. The model proposed is trained and tested by using the actual transmission line project data. The prediction results show that the XGBoost model has a great advantage over neural network and support vector machine (SVM) model in terms of prediction accuracy and deviation of results, and can output the index importance ranking, which can provide effective investment and control index reference for decision makers. And the model is highly reliable and interpretable.