固定影响平行数据模型参数的偏最小二乘估计及应用

PLS estimation and application of fixed effect panel data model parameters

当平行数据的解释变量之间存在严重的多重共线性,或者样本点个数与解释变量个数相比较少时,用经典方法求解模型(如最小二乘法等)误差偏大,难以满足实际要求。针对这种情况提出了用偏最小二乘法思想求解固定影响平行数据模型,并且实例表明误差明显减少,可以满足建模和预报的要求。

when the explanatory variables of panel data exist serious multicollinearity,or the number of Samples is very small compared with the number of explanatory variables,we can＇t use classical methods（such as method of least minimum square）.Otherwise the error is great enough,which can＇t satisfy actual requirement.Because of this complexion,this article brings forward PLS estimation to solve the fixed effect panel data model,which makes the error decreasing,and it can satisfy the demand of modeling and prediction.