Article source: data kaleidoscope

Click on the blue letters above to follow us

The structure of the table of contents is as follows:

1. Three major analytical thoughts

2. Analysis methods of different life cycles

3. Practical case summary

 

To master the common data analysis methodology is the basis of cultivating data analysis thinking. As the saying goes, “To do a good job, you must sharpen your tools first”, and data analysis methodology is one of the most powerful weapons of data analysis. This section will focus on the common analysis methods of data analysis and generally introduce the data analysis methodology commonly used by data analysts in their daily work.

01

Three major analytical thoughts

Comparative analysis, user clustering and correlation and causality analysis are the three major analytical thinking throughout the whole process of data analysis. As shown in Figure 1, the three major analytical thoughts contain different analytical methods.

FIG. 1 Generalizations of three major analytical thoughts

There is no clear data conclusion without comparison. Comparative analysis can measure the overall size of data, data fluctuation and data change trend. Therefore, comparative analysis is the simplest and most effective method to draw data conclusions. Typically, data analysts compare business data with broader market data or industry gold standard data to determine business status. In addition, year-on-year, sequential/horizontal/vertical ratio are also commonly used comparative analysis methods. A/B test is A special comparative analysis method, which is the online test method commonly used by data analysts and the most effective method to explore the causal relationship between variables.

User clustering is also the analytical thinking that runs through the whole data analysis link. User clustering based on the characteristics of user behavior data/consumption data is the basis of user refined operation. Users can be grouped based on the historical data of users, and the data can be divided into boxes to form labels of regular types, so that users can be grouped according to labels. If the enterprise’s data label system can be done well, users can be grouped directly through data labels. In addition, user cohort analysis is another method of user cohort classification. This method is a combination of horizontal and vertical analysis method, which analyzes the changes of cohort as the cycle progresses horizontally and analyzes the differences between groups at the same stage of life cycle vertically. Of course, data analysts can also use the RFM model or machine learning algorithms such as K-means to achieve user clustering as needed.

In addition to comparative analysis and user clustering, correlation and causality analysis is the third major analytical thinking that data analysts need to possess. In the process of exploring the relationship between variables, correlation analysts often use analysis methods, but the correlation between variables does not mean that they have causality. Therefore, causal inference is also the analysis method that data analysts must use when necessary.

Full article please move, data kaleidoscope

The remaining directory structure is:

2. Analysis methods of different life cycles

3. Practical case summary