0 1. Descriptive statistics

Descriptive statistics is a method of sorting out and analyzing data through charts or mathematical methods, and estimating and describing the relationship between data distribution, digital characteristics and random variables. Descriptive statistics can be divided into three parts: centralized trend analysis, decentralized trend analysis and correlation analysis.

Centralized trend analysis

Centralized trend analysis mainly relies on statistical indicators such as mean, median and mode to express the centralized trend of data. For example, the average score of the subjects is more or less positive or negative.

Deviation trend analysis

Deviation trend analysis mainly relies on full scale, quartile deviation, average difference, variance (covariance: statistics used to measure the relationship between two random variables), standard deviation and other statistical indicators to study the deviation trend of data. For example, if we want to know which of the two classes' Chinese scores is more scattered, we can compare them by four points or several percentage points.

correlation analysis

Correlation analysis discusses whether there is statistical correlation between data. This relationship includes not only a single correlation between two data, such as the relationship between age and personal domain space, but also multiple correlations between multiple data, such as the relationship between age, incidence of depression and personal domain space. It includes both the linear correlation of A big and B big (small) and the complex correlation (A = Y-B * X). It can be that the A and B variables increase this positive correlation at the same time, or that the B variable decreases this negative correlation when the A variable increases, and it also includes the closeness of the two variables * * * with the change, that is, the correlation coefficient.

In fact, the only data relationship that has not been studied is the internal basis of data covariation, that is, causality. What is the use of obtaining correlation coefficient? In short, with the correlation coefficient, we can estimate the variables from A to B according to the regression equation, which is regression analysis. Therefore, correlation analysis is a complete statistical research method, which runs through hypothesis, data research, data analysis and data research.

For example, we want to know what changes can be made in prison scenes to reduce prisoners' violent tendencies. We need to arrange and combine different cell colors, cell greening degree, cell population density, outdoor time and visiting time, and then make each cell undergo an experimental treatment, and then use factor analysis to find out the factor with the highest correlation coefficient with prisoners' violent tendency. Assuming that this factor is cell population density, we randomly divide the subjects into more than a dozen cells with different population densities, and then get two groups of variables (namely, variables A and B) of population density and violence tendency. Then, we put the population density on the X axis and the violent tendency on the Y axis, and get a valuable chart. When a warden wants to know how much the tendency to violence can be reduced when a cell is expanded to n people/cell. We can bring the current population density and the reconstructed population density into the corresponding regression equation and calculate the expected violent tendency before and after expansion. The difference between the two data is what the warden wants to know.