Nonparametric Statistical Inference by method of Local Moments

数据压缩是由DuMouchel等人提出的。艾尔(1999)处理海量数据集。其想法是将数据集缩小到更小的代表性样本--“挤压数据”，而不是将算法扩大到大型数据集。然而，原始的数据压缩方案可能在计算上仍然是繁重的，因为它通常需要求解一个大的方程组。在这个项目中，我们开发了一种新的数据压缩方法，该方法不需要解决应用于压缩数据的极大似然估计的理论性质，并提供了一些仿真研究来解释我们新方法的有效性。

The data squashing is proposed by DuMouchel, et. al. (1999) to deal with massive data sets. The idea is to scale data sets down to smaller representative samples, "squashed data", instead of scaling up algorithms to large data sets. However, the original scheme for the data squashing may still be computationally burdensome because it usually requires solving a large system of equations. In this project, we developed a new method for the data squashing which does not involve solving theoretical properties of the maximum likelihood estimator applied to the squashed data.Some simulation study was provided to explain some evidence of effectiveness of our new method.