Abstract: In order to predict the gas emission in the mine coal face effectively, combining the various factors which affect the gas emission, for the serious multicollinearity among factors causing the amplification of calculating error, it uses stepwise regression algorithm to optimize. Through SPSS software, it integrates multivariate principal component regression analysis algorithm to get the main factors which affect coal face gas emission, and establishes PCR-SPA predictive model. It compares the forecasting performance of the model with multiple linear regression prediction, gray correlation analysis prediction, BP neural network prediction results. The results show that：the model selects three main components variables, effectively reducing the amount of computation, with a high prediction accuracy, which can predict the mine face gas emission effectively.

Key words: coal working face, Statistical Package for Social Science（SPSS）, multicollinearity, principal component analysis, gas emission quantity

摘要： 为了对矿井采煤工作面瓦斯涌出量进行有效预测，结合影响工作面瓦斯涌出量的各个因素，针对各因素间存在严重的多重共线性引起算法计算误差放大的问题，采用逐步回归算法进行优化。运用SPSS软件，综合多元主成分回归分析算法得出采煤工作面瓦斯涌出的主要影响因素，并建立PCR-SPA预测模型。将该模型的预测性能与多元线性回归预测、灰色关联度分析预测、BP神经网络预测结果进行对比分析。结果表明：该模型选取了三个主成分变量，有效地减少了计算量，具有较高的预测精度，可以对矿井工作面瓦斯涌出量进行有效预测。

关键词: 采煤工作面, 多元统计分析软件（SPSS）, 多重共线性, 主成分分析, 瓦斯涌出量