The previous "Know these 7 things before data analysis, spend 80% less time" talks about what you need to know before data analysis. When entering the data analysis stage, the author summarizes and organizes some easy-to-use and easy-to-use analysis methods, so that you can do more with less . Below, Enjoy~ 1. Pyramid Model of Data Analysis Easy-to-use business email list data analysis methods for product managers Data analysis can be roughly divided into five levels in terms of difficulty, covering the process of data sorting, statistics, and machine learning. The decision-making link after data analysis is not within the scope of this discussion.
The five levels include: quantifying the current situation, quantifying the correlation between variables, quantifying the causal relationship between variables, predicting the future and finding the best solution. Q1: What is the trend of the number of visits to new customers by SaaS salespeople? For example, at level 1, the problem that needs to be solved is to business email list quantify the current situation. You can use "narrative statistics", "data visualization" and other methods to express the data through charts, and observe the number, frequency, fluctuation, distribution, etc. of the charts. This level is mainly realized by relying on the observation after visualization. Q2: Is the number of visits to new customers by SaaS salespersons related to turnover? Q3.
Can the marketing department increase the amount of advertising to get more orders? For example, exploring the correlation and causality of data in levels 2 and 3 requires secondary inference after data observation, which belongs to data mining. Q4: What data does the SaaS product manager track to know if the user will cancel the subscription next month? Q5: How should the CEO allocate employees to develop new customers to maximize revenue? Levels 4 and 5 hope to use samples of existing data to estimate the possibility/probability, optimal solution or approximate solution of future data, which belongs to data exploration. For different levels, the problems that need to be faced are different. You can progress layer by layer, from easy to difficult, and select the problems that need to be solved for data analysis one by one.