At noon today, Mr. Oil Painting has not been taken to pick up the smoke and chat with the lawn after a meal. Sitting in the office, actually summarizing the statistical approach. :)
Statistical analysis method
Modern statistical analysis method
Classification analysis
Cluster analysis
Discriminant analysis
Qualitative data analysis
Structural simplification analysis
Return
Cluster analysis
Principal component analysis
factor analysis
Correspondence
Related analysis method
Qualitative data analysis
regression analysis
Typical related analysis
Principal component analysis
factor analysis
Correspondence
Prediction decision method
regression analysis
Discriminant analysis
Qualitative data analysis
Cluster analysis
Statistical analysis of a qualitative variable
Multiple distribution and X inspection
List analysis
Consistency test
Fitting optimization test
Second, the regression analysis
Overview: The number of variables between the variables, determines the mathematical relationship between the argument and the variable, establish a regression equation, and perform various statistical inspections on the regression equation, and can be predicted.
Third, mean analysis
Overview: The mean analysis process calculates a comprehensive description statistic of the specified variable, including the analysis of the overall characteristics and the discrete trend. The analysis includes statistic mean, sum, and two of the discrete trends include the minimum statistic, maximum, standard deviation, variance, very difference and mean standards.
Fourth, frequency analysis
SUMMARY: The data is summarized by group, forming a frequency analysis table of variables, forming a general understanding of the number of characteristics of data and the distribution of data.
V. Description analysis
Packet statistical analysis of data, resulting in the aggregation of strain, discrete trend, abnormal value of each indicator, proportional table, stem graph, box sharing map.
Sixth, related analysis
Related analysis is a method of analyzing the number of dependencies of objective things, which primarily portrays two types of variable linearly related closures, and two variables x, Y is all random variables, and is equally position. The relationship between the two variables can be reflected by drawing a scatter plot or a calculated relationship coefficient. This section includes two-dollar variable correlation analysis and partial correlation analysis.
Seven, variance analysis
A single variable variance analysis is performed, that is, the overalls of two and more than two or more independent samples from mean are the same. This process requires the analysis variable as a normal distribution, otherwise the card party inspection or a variety of features included in this inspection should be selected.
Eight, multi-sample inspection
It is a significant difference in the distribution of multiple independent overalls from the sample from the sample. Its method is to infer whether there is a significant difference in the mean or the median number of independent overalls. A multi-sample test provides two commonly used test methods: KruskAlwall H and median testing.
Nine, one yuan linear regression
slightly. Formula, this gentleman is not written.
Ten, multi linear regression
slightly. Formula, this gentleman is not written.
Eleven, gradually return
Variable screening is performed based on pre-set selected and eliminated criteria. First, the respective variables are calculated, respectively, and the first entry equation that is maximized by large to small selection; then recalculates the contribution of the respective variables, and inspects whether the variables already in the equation is due to the introduction of new variables. No more statistical significance. If yes, it is removed and recalculate the contribution of each variable to Y. If there is still a variable below the selected criteria, continue to consider, until there is no variables in the equation that can be removed, and there is no variable outside the equation can be introduced.
Twelve, cluster analysis
Distance and similarity factor
System clustering method
Thirteen, Fuzzy cluster analysis
Overview: Cluster analysis is a method of studying things classification, where the classification object is divided into several classes in accordance with certain rules, but is determined based on the characteristics of data. Popular lecture, cluster analysis is used to simplify the data, that is, the similar individual (observation) is at a group.
14, factor analysis
Factor analysis is the promotion of primary component analysis, and it is also a pluralistic analysis method for multiple variables into few comprehensive variables.
