Correlation analysis, specifically: finding the correlation between two data indicators. For example, in an APP, the user repeatedly browses a product, so will he buy it? One faction believes that if you watch a lot, it means that users are interested, so they will buy it The other school believes that if you don’t buy it after watching it for so long, you will definitely not buy it. There is also a school of thought: It doesn't matter how many times you watch it, it depends on whether there is an event or not. It seems reasonable to listen to, but in the end, the data must speak. What is discussed here is whether there is a relationship between user browsing behavior and consumption behavior.
Correlation analysis is to find out the relationship between these two indicators. 3. Directly related relationship Note: There may be inherent correlations between indicators. There are three common forms: In the structural analysis method, the relationship between the overall index and the partial index In the index disassembly method, the relationship mobile number list between the main index and the sub-index In the funnel analysis method, the relationship between the before and after step indicators (As shown below) Nine Data Analysis Methods: Correlation Analysis These three situations are called: direct correlation and direct correlation do not require data calculation, and the relationship can be clearly seen through index combing. In the case of direct correlation, it is well understood that the two indicators have a simultaneous upward/downward trend. for example.
The performance of the entire company is not good, so the performance of branch A is also very poor (structural analysis) The number of customers is too small, so the overall performance is not good (main indicator, sub-indicator) The number of people who saw the ad was too small, so the final conversion was not good (before and after steps) If two indicators that are directly related do not rise and fall at the same time, it often means problems. For example, with user growth, the number of new registered users has risen sharply, but the paid conversion rate has continued to drop sharply, which means that the efficiency of customer acquisition is declining. It may be that the target users have been exhausted, the channel may be fraudulent, or it may be that the acquisition The customer method is wrong, in short, an in-depth analysis is required (as shown in the figure below).