Thinking: Data Analysis and Data Background Design

Many years later, when faced with many numbers, I will definitely be reminded of that distant summer when my teacher taught me hypothesis testing.

I imitated "One Hundred Years of Solitude" and recalled the course of studying experimental statistical design in college. At that time, I didn't care about the boring courses such as statistics and experimental design. Don't make up for the knowledge of doing experiments, monitoring data, observation and analysis, and analyzing data. After writing the thesis, I sighed that I finally got out of the bitter sea.

But there are always many reincarnations like "fate" in life. When I graduated, I thought I was out of the misery and no longer had to deal with mobile number list boring numbers and statistics. However, the numbers after work are still something I can't live without. Every time the developers questioned me and asked me to come up with data and analysis conclusions to prove my point of view and demand reliability, I was like Colonel Aureliano in "One Hundred Years of Solitude", and I fell into recalling and doubting the choice of the past.

Remember to keep an eye on the experimental data every time when doing graduation design .

Experiments, for fear of abnormal data fluctuation experiments. After the experiment is over, the collected data is sorted out, and every time a long list of inter-day data is used to think about whether there is any problem, and finally, the manual Excel "data background" is used to conduct in-depth analysis and sorting. Also remember to use the hypothesis testing procedure, in which the significance of the conclusion is demonstrated with an orthogonal test method. Reluctantly completed a seemingly scientific thesis, just graduated in such a confused way.

A few years later, when I was frustrated at work, I remembered that I had done data analysis in such a "professional" way, but now I can only sigh in the face of the data.

So I thought about sharing some ideas about data analysis and designing the data backend through this little story of my graduation thesis, without talking about specific methods, sharing some experiences from the perspective of thinking, and helping you find ideas and directions for solving problems Be inspired.

First, let’s talk about the direction of data analysis.

I divide data analysis into three parts: monitoring, observation, and analysis in the order of execution. It can be understood that monitoring data is the basis of observation data, observation data is the source of analysis data, and analysis data is the result of a data analysis behavior. So let's start with monitoring data.