中文摘要：

针对包含饱和样本数据的频数幂律分布拟合, 提出一个新的幂律分布指数的极大似然估计方法的修正公式. 对比研究显示, 修正公式适用于剔除异常饱和值的 幂律频数拟合. 如果不剔除饱和值, 幂律指数的估计只能使用修正前的公式, 其误差随幂律指数变化, 指数较小时误差较大. 由此建议, 对于包含饱和样本的频数分布拟合, 首先剔除异常的饱和值, 然后对剩余不含饱和值的子集使用修正公式进行参数估计.

英文摘要：

We developed a new correction formula for fitting the frequency distribution with saturation data. Through comparing four cases （with/without the saturation data and with/without the modified formula） to fit the tested numerical data, we found that the correction formula is the best for the data excluding the saturation data, although the original MLE （Maximum Likelihood Estimator） may be acceptable for the data including the saturation data and for a larger index. We therefore suggest to use the modified MLE to fit power-law frequency distribution, and the saturation data should be rejected first.