1. Background
This article mainly shares how to do user portraits in some small and medium-sized companies, but the portrait data is incomplete. Everyone knows that for example: in Alipay or some banks, there are your ID card data and some deposit data. Get your age and your assets, but what about other companies? For example, I still want to know the asset situation or consumption situation, but if it is not for some financial companies, such as game companies, video companies, and even some smaller companies, they will not be able to get user asset data at all.
Therefore, is it possible to consider using a compromise method for asset evaluation, for example: what is the situation of the user's mobile phone? For example: Is it an Apple phone or a Redmi phone, or is it the latest iPhone or the iPhone 4? Generally, as long as an app or website is installed or visited, the user's mobile device information can be collected.
2. Data processing ideas
When a user visits e-commerce websites such as JD.com or Taobao or plays King of Glory or even watches videos, he will report device-related information, such as device model information, storage information, etc. The data is shown in the table below, but the actual If the data such as mobile phone brand and model cannot be obtained, it can only be obtained after processing.
The specific processing flow is as follows:
Raw data collected
Step 1 : You need to map the mobile phone model to the marketing name. For example: iPhone3,1 actually corresponds to iphone 4; M1805E2A corresponds to Xiaomi 8SE. For specific mapping website reference: https://github.com/KHwang9883/MobileModels , https://blog.csdn.net/qq_19926599/article/details/86747401?spm=1001.2014.3001.5506
Step 2 : Crawl the information data of Zhongguancun mobile phone and get the following table
Step 3 : Finally, we can match users with the help of the dictionary mapping data in the first step
In this way, the detailed data of a device's mobile phone is obtained. The crawler code reference: https://github.com/zhugezifang/webmagic-collector
3. Summary
When some small and medium-sized companies do not have complete data, they can consider using existing data and combining some data that can be extracted, so that some valuable data can also be processed. For example: you can do some marketing based on the price, brand, or even series of processed mobile phones, and place higher-quality advertisements for high-priced mobile phones or high-end phones (high-end phones from Apple and Huawei)
This article only uses mobile phone information as an example to explain how to make user portraits in small and medium-sized companies. Many small and medium-sized companies do not have as complete data as large companies, resulting in very limited data that can be mined. This article also provides an idea how to use limited data. In some cases, the combination of existing data and online data fusion can achieve the desired requirements.