Big data is the collection of huge amounts of data, the Internet, the Internet of Things, wearable devices, etc., in this era of the Internet, people leave the behavior of the data is recorded all the time, creating a huge amount of data, which in turn appeared in the emergence of big data analysis and mining and other positions. Through the analysis of big data mining, you can find historical patterns and predictions for the future, which is the core goal of big data analysis.

So if you use big data to dig deep into the Internet potential users? The following is an introduction to the business logic process.

1, the potential target user portrait

First you need to characterize your potential user groups, including: user groups are mainly active in which channels, *** with the same characteristics (preferences, occupation, income, spending power, etc.) what, etc., through the user portrait, a clear understanding of what you want to mine some of the characteristics of the potential user groups and the law of activities, so as to dig model The data source and conditions to support the mining model.

For example, if you need to tap into the potential users of decoration, their active channels are mainly in the major home furnishing websites, home network, decoration design network and other house-related websites or apps, and generally these users will browse these websites/apps in advance to make preparations.

2, data collection

After clarifying the active channels of potential users, you can target the collection of data, data collection needs to be cleaned, converted, loaded, and some of the useless data will be screened in advance to ensure the quality of data.

3, data modeling

This stage is very important, through the data modeling to analyze the potential target users, modeling is a very complex work, the need for user behavioral data, portrait data to split, merge, correlation, so as to establish a set of data models or more.

Taking decoration as an example:

(1) Consumption ability model, we can analyze the consumption ability of users based on the price of the furniture they browse, previous consumption history, income, etc.

(2) Quality customer analysis model, we can analyze the data based on the number of times a user browses, length of stay, purchase records, credibility, etc., so as to derive the user's decoration urgency, which can be categorized into high, medium and low levels.

At the same time, you can also join the user's region, district and other dimensions of the analysis (according to the specific needs), the model will be subdivided, and finally can be associated with the collision through the various models, combined into a variety of models, such as strong spending power and immediately want to decorate the potential users, strong spending power is not too urgent to require the decoration of the potential users, etc., so that you can realize the differentiation and precision of the operation. (Example of a very simple, in fact, really do it is still very complex, all factors should be taken into account)

4, the development of verification

Data modeling is completed, it is necessary to complete the research and development and used in combat, to test the accuracy of the data model in the end how to go according to the results of the model to make adjustments.

Big data analytics is supposed to be a prediction of what is going to happen in the future, and this kind of uncertain prediction is constantly changing with the development of society, time, place, environment, policies, etc. So when we do analysis and mining, we need to quickly and constantly adjust by trial and error, so as to achieve a more accurate analysis results.