Data distribution mainly depends on geometric distribution, Poisson distribution and binomial distribution to study the distribution trend of data. For example, is the overall distribution of the target data segment divergent or centralized? What frequency band is it on? In which range is the median concentrated? In which data range is 80% of the data concentrated? The purpose of looking at the distribution is to understand whether the business data is stable and the concentration of the data.
2. Normal distribution
Normal data types can be divided into continuous data and discrete data according to attributes. Continuous data belongs to data that can be subdivided continuously, such as length, width, height, density, temperature and so on. Discrete data can't be subdivided, mainly expressing the attributes of objective things, such as numbers, attributes, ratios, etc.
3. Statistical sampling
Statistical sampling involves how to design samples, point estimators and proportional sampling analysis. It is difficult to analyze massive data and check data distribution. It is necessary to sample the samples and reflect the distribution of the whole sample through the distribution of the samples.