How to do data analysis well? There are 5 key links!

In the era of big data and mobile Internet , everyone who uses a mobile terminal is producing data all the time, and products provided by Internet services are also continuously accumulating data. Data, like artificial intelligence , often shows a more objective and rational side. Data can help people understand the world more intuitively and clearly, and data can also guide people to make more rational decisions.

In the data-driven era, no matter what your job is, mastering certain data analysis capabilities can help you better understand the world and improve your work efficiency.

The data analysis process is mainly divided into five key links: clarifying the purpose of data analysis , clarifying the data source and data caliber , data processing , data analysis , and output .

1. Clarify the purpose of data analysis

Everything has a purpose before it is done, and the same is true for data analysis. Before conducting data analysis, we must first clarify why we need to do data analysis? The purpose of data analysis is clarified below by breaking it down into three elements: users , needs , and scenarios .

1. User

The users mentioned here refer to who will see the data analysis content or results? The target users here are mainly divided into three categories: yourself, internal business departments of the enterprise, and external customers. Here we mainly analyze the latter two.

Internal business departments:

This type of user usually develops different strategies to improve certain indicators of the company. They can be the marketing department, operations department or maintenance department. They often lead companies to accumulate a large amount of data, but they don’t know how to use it or how to make effective decisions through data analysis.

External Customers:

This type of user usually does not have industry data in a certain field or multiple fields, and hopes to understand his users or the market through this data. Your company happens to have such data. In this case, by realizing the value of the data and forming external data analysis outputs, external users can better understand the market, and you can also realize the value of the data and bring profits to the company.

2. Demand

Why do your users, i.e. the data analysis question raisers, want to do data analysis? Do they hope to discover problems through data analysis or do they hope to improve certain business indicators? These are all things you need to understand before doing data analysis. Only by understanding the needs can you develop more reasonable data analysis ideas (data analysis methods will be introduced later).

3. Scene

Scenarios more often reflect data analysis scenarios. For example, if the business department wants to understand the reasons for user churn during the registration process, then this is the scenario of the problem. It is necessary to define the problem based on the scenario, sort out data analysis ideas, and choose data analysis methods.

2. Clarify data sources and data caliber

1. Data Source

There are three main ways to acquire data. The first is to acquire data through some data collection tools based on the front-end page, such as visual data collection products such as GrowingIO. The second is to use data embedding during the product design process, which allows for simple extraction when data is needed. The premise of this method is that preparations for future data acquisition have been made in advance during the product planning stage. The third is to find the R&D team to acquire data through background scripts or technical development if no functional embedding is performed in the early stage and visual collection tools are unable to acquire data.

2. Data caliber

Data caliber refers to the definition of the meaning of a certain data indicator. To give a simple example, the definition of user churn indicators may vary for different products or different fields. For ordinary e-commerce products, users are considered lost if they do not log in or make a purchase within three days. However, for luxury e-commerce, it is unreasonable to count users as lost if they do not log in or make a purchase for just a few days.

To clarify the data caliber, it is necessary to combine the needs of the person proposing the data analysis task and the specific business scenario. Defining a clear data caliber is of key significance to subsequent data processing and data analysis.

3. Data Processing

The main tasks in the data processing stage are data cleaning, data completion, and data integration.

1. Data cleaning

Discover outliers in the data. For example, when processing user login data for multiple consecutive days, if the number of logins on one day far exceeds the normal value, then it is necessary to analyze whether there is a major marketing activity on that day or an error occurred when collecting data. Outliers can not only discover problems with data collection methods, but also find the target of data analysis through outliers. For example, the analysis of credit card fraud is done by looking for abnormal data.

2. Data completion

How to solve the problem of missing data? One way is to fill in the average value based on the correlation between the data before and after, and the other is to directly choose to lose the record and not use it for data analysis. Both methods have their own advantages and disadvantages. It is recommended to analyze them based on specific problems.

3. Data Integration

When collecting data, there may be potential correlations between different types of data. By integrating the data and enriching the data dimensions, it is more conducive to discovering more valuable information. For example, if user registration data is associated with user purchase data, the user's basic attribute information can be used to determine whether the product purchased by the user is for personal use or as a gift.

4. Data Analysis

Data analysis ideas are also called data analysis methods. Data analysis must be purpose-oriented, and the data analysis method is selected based on the purpose. Generally speaking, there are the following main analysis ideas

1. Abnormal analysis

Discover abnormal situations through data analysis and find ways to solve them.

2. Find the correlation

Association relationships can also become shopping cart analysis. The well-known case of Walmart diapers and beer is the best practice of association relationships. By analyzing the relationship between different products or different behaviors, user habits can be discovered.

3. Classification and stratification

Users are classified and stratified based on their characteristics and behaviors to form refined operations and precise business recommendations, further improving operational efficiency and conversion rates .

4. Prediction

Predict users' possible future behaviors through their historical behaviors, and improve user perception and usage experience.

5. Output

As mentioned above, the purpose of data analysis is to clearly understand users, products and current business forms through data, so as to obtain effective strategic decisions to guide the next step of development.

How to clearly understand users, products and business ecosystems through data? Rows of boring numbers cannot allow business departments or external customers to intuitively understand the meaning behind the data, so data visualization methods are needed. Simply put, rows of data are converted into charts to intuitively show data trends, correlations between data, etc. When visualizing data, it is important to consider how many dimensions the data has and what the data should be presented to the viewer. These all affect the form of data visualization.

For example, a pie chart can be used to show the gender ratio of registered users, a curve chart can be used to show the growth trend of the number of registered users over time, and a bar chart or map can be used to show the location of registered users, etc. When choosing a visualization method, you must fully consider the characteristics of the data and what you want to express with the chart, so that you can present more intuitive analysis results in a reasonable way.

In addition, the output of data analysis is usually presented in the form of a data analysis report. The main structure of the data analysis report is as follows:

1. Data analysis background
2. Data source and data description
3. Data analysis methods
4. Data Visualization
5. Data Decision Making

The above is the general framework of a more formal data analysis report. If the daily report does not require a formal presentation of data analysis results, specific analysis can be done on specific issues.

VI. Conclusion

Data analysis methodology must serve to guide specific work practices, so it is not enough to just master the methodology, but also to continuously improve and optimize the methods through practice. Only when you actually do data analysis can you discover your own shortcomings. No matter how much you say, it is better to just do it.

The author of this article @静静 is compiled and published by (Qinggua Media). Please indicate the author information and source when reprinting!

Product promotion services: APP promotion services, advertising platform, Longyou Games