(Data analysis) Three core thinking in data analysis

1. Structured

Structured: structure analytical thinking

• Summarize and organize the arguments
• Progress and disassemble the argument
• Complete and supplement the argument

Structural elements

Core argument: Look for the top of the pyramid. It can be hypothesis, problem, prediction, cause.
Structural dismantling: from top to bottom, the core arguments are dismantled layer by layer into constituent arguments, and there is a causal or dependent relationship between the top and bottom.
MECE: Independent and completely exhausted. Avoid overlapping and duplication between arguments, and sub-arguments should be as perfect as possible.
Verification: Regardless of the core arguments or sub-arguments, they should be quantifiable. From the data point of view, they must be verifiable.

Example:
Problem: There is a product that is sold offline, and suddenly I find that a certain month's sales decline, analyze the reasons for the decline.

First find the core argument, that is, the decline in sales. Disassemble the core argument into two sub-themes: internal factors and external factors. The sub-thesis will be further disassembled. In the process of dismantling, follow the structural elements. Then you can get the following simple mind map. The argument in the picture is simply split, and you can continue to disassemble it according to the situation.

2. Formulation

Structured cannot be directly used for data analysis. Data analysis is based on the corresponding analysis of the data. In the process of structured, there is no accurate data as support, and it has the characteristic of divergence. When we are structuring, many arguments may be just conjectures, and these conjectures may not be verified by data. So many times we also use formula.

Formula: All knots can be quantified, and the final argument after disassembly must satisfy: the principle of least indivisible.

Still use the above example:
we formulate it simply, and the result is shown in the figure.

Each argument in the figure can be obtained through data analysis or existing data.

3. Commercialization

Commercialization: Complement the shortcomings of structured and formulation.

Question: How to estimate the number of shared bicycles in a city

• Floating population from cities
• Calculated from population density
• Calculate from city traffic data
• Calculated from bicycle ownership

Bicycles are lossy, and the consumption of bicycles should be considered in the calculation formula.

Using structured thinking and formulating dismantling, the final analysis arguments obtained are often phenomena. But it does not mean that it is the cause of the problem. At this time, we need to use business thinking to further explore.

Knowledge reference: data analysis tutorial