基于Xgboost算法的国际期货涨跌预测分析

Analysis of the Rise and Fall of International Futures Based on Xgboost Algorithm

摘要: 基于高效复杂的xgboost算法构建了分类预测模型，对最近三年国际期货的日交易数据进行了训练测试。该模型通过调参工具遍历所有参数组合，得出最优参数。然后，对比于决策树、随机森林、支持向量机算法，结合多个评价指标进行综合评价。实验表明，xgboost算法构建的模型各项指标均高于其他算法，综合预测能力更好。同时，也为期货价格预测提供了一种有效的新方法。

Abstract: Based on an efficient and complex xgboost algorithm, a classification prediction model was con-structed to train and test the daily transaction data of international futures in the past three years. The model traverses all parameter combinations through the reference tool to obtain the optimal parameters. Then, it is compared with decision tree, random forest, support vector machine algo-rithm, and combined with multiple evaluation indicators for comprehensive evaluation. Experi-ments show that the indicators of the model constructed by the xgboost algorithm are higher than other algorithms, and the overall prediction ability is better. At the same time, it also provides an effective new method for forecasting futures prices.