Distribution Fitting Based On Certain Complicated Observation Data

Data processing and Data analysis exist in various fields of science research and actual production. It is very valuable and significative to analyze the distribution of complicated data, and the result is great important to actual application, but the distribution is hardily confirmed.In this paper, the approximate distribution fitting based on certain complicated data is researched. The following issues are investigated:1. The general method of analyzing complicated data distribution is systemically explained. The principle and specific steps of chi-square test and kolmogorov test are expounded.2. By analyzing this complicated data and data after the logarithm based on different grouping, it showed that this data does not obey normal distribution or lognormal distribution. According to grouping, the most superior one grouping is selected. Because the data histogram displays the obvious dissymmetry, the fitting with the segment function is proposed.3. Analyzing the characteristic of fitted points, some fitting functions are constructed. According to the Gauss-Newton algorithm, parameters are estimated. Then fitting result of different function is appraised through curve graph and two kinds of tests.4. Select the best fitting segment function, normalize it, the density function is obtained. Finally, the quantile table of the approximate is provided.