Data exploration of machine learning — data feature analysis (Pareto analysis)

Time:2020-6-6

Pareto analysis, or contribution analysis, looks for a small number of key or decisive factors in all factors.

There is a common law, the 28 law. All walks of life are interpreting its profound meaning:
For example:
1. 80% of the company’s profits come from 20% of the best-selling products, while the other 80% of the products only generate 20% of the profits;
2 about 80% of the world’s resources are exhausted by 20% of the world’s population;
3 80% of the world’s wealth is owned by 20% of people;
20% of the population or 20% of the diseases consume 80% of the medical resources.
In some special fields, the law of 28 may be derived from the law of 19 or even more.

In the aspect of data mining, it is also necessary to find the independent variable elements that affect the dependent variable to the greatest extent according to the idea of 28 law.

As you can see in the above example, the profit share of the first seven items has reached 80%.
Especially in commodity sales, the results of Pareto analysis can be used to guide businesses to focus on key products in commodity building.