数据分析工作流程

Data analysis, as the name implies, can only be analyzed if there is data. No matter what analysis you do, getting data is always the most important thing. A skillful wife cannot make bricks without rice. Do as much data as possible, otherwise everything is out of the question.

There are many kinds of data, of course, we can't grasp them all at once, first of all, we should make clear what kind of data we need, that is, what the goal is.

The data that data analysis focuses on is generally oriented to business topics, such as what data is in the product, operation, sales, marketing and other business sectors of data services, and what data we need. These are what a qualified data analyst must sort out. Specifically, it can be divided into two parts:

At this stage, at least I can understand the data on hand, understand where each data is used, and what is the point.

With grain in hand, the data model began to be built! Build the data model out loud, take it out and ask others what it is, what you want to express, what indicators are normal and what is abnormal. It would be strange not to be asked with a silly face.

Not all indicators will be given by the business side. It is quite possible that some data require baselines based on experience, common sense and analysis.

After crossing the second step, as a data analyst, you already have a good understanding of the common trends of the business. Moreover, if you have the ability to make a judgment, you will be able to identify the abnormal state. Not all anomalies are uncontrollable, and a considerable number of anomalies are caused by observable factors, such as the macro environment, such as marketing actions and promotion plans. At this point, it is necessary to have the ability to understand these clearly identifiable "white rhinos" before considering other "black swan" problems.

Therefore, the task of this step is to disassemble the factors that affect the business so that they can be quantified, represented by one or more data indicators, and can be analyzed in combination with internal data to this extent. That is, * disassembly factor *

Specific how to dismantle, is the need for a lot of analytical work to do the basis, and finally precipitated is the most valuable experience.

After completing the third step, the data analyst can judge the obvious and significant influencing factors. If you want to further improve, you have to go deep into the details to understand how the activity is done, how the product is designed, and what users need. These have reached the core of the business, so it is difficult to figure it out in a short time and needs to be accumulated over a long period of time.

The more you get to the core of the business, the less continuous indicators can describe and the more you need to label them. Similar user emotions, preferences, product function points, marketing logic, are described and quantified by tags. It's not like dragging out the existing dimension tables in the database. How to label accurately, quantify the business, and make a clear classification is the skill of senior data analysts.

But! In real work, quite a lot of problems come not from technology, but from business.

The purpose of doing this step is not only to precipitate the experience and improve the pressure, but also to transfer knowledge and expand the scope of service. It is difficult to guarantee that a person who has mastered excellent analytical skills will be an expert in a line of work without changing the profession all his life. The transfer of knowledge depends on the summary of the model and the accumulation of experience, both of which are indispensable.

The more I know about data analysis, the more I feel like this profession. This position requires not only that people have the skills to solve problems by themselves, but also that they understand the business, the process and why they do it. This is the complex of Tao, dharma and art.

At the enterprise level, the mission of data analysts is not only to collect data, do processing and analysis, but also to provide decision support for management and contribute to the development and growth of enterprises.