As a data analyst in a business department, the scenario that is often faced is to analyze anomalies. For example, the boss finds you and says, "Guy, help me analyze the reason for the decline in GMV in the last week." When many people get this question, they start to look for various dimensions to split it, but they are still at a loss in the end. When we first get this business email list question, we need to use the analytical method, the description and the comparative analysis method. It is to use a descriptive analysis method to see how much the specific GMV value has dropped, and to use the trend and comparison method to judge whether it is a real abnormality.
It is possible that the same period last year was a normal decline after the holidays. This process is like knowing some basic information about the crime scene when a detective detects a case. After having a basic understanding of the anomaly, if you want to understand the truth, you need a subdivided thinking method, and step by step to business email list decompose and restore the clues of the incident. But how to subdivide, and according to what kind of routine subdivision is exquisite. First of all, the common methods of subdivision are introduced, and the author summarizes them into three categories. Usually in the analysis of a specific problem may be used together.
Structural subdivision The first category is broken down by structure: the structure of the thing or organization itself; The distinction of production lines (such as cosmetics, fresh food, etc.), the distinction of regions (such as East China/West China/South China), the distinction of channels (APP/mini program/PC terminal), etc. The DuPont analysis method, invented by DuPont, adopts a pyramid structure and decomposes the return on equity of the enterprise into the ratio or product of multiple financial indicators step by step. This is borrowed from the analysis of our conventional indicator system. For example, in the scenario encountered at the beginning of the article,.