If you have several years of experience in data analysis, or have some statistical algorithms/machine learning skills, it is recommended that you master “web mining” in order to improve your personal ability.

But many people know little about “network mining”, what is network mining?

Network (hereinafter referred to as the network data mining mining) is a mainstream, important data mining technology, common, such as social networking, shopping networks, financial networks and network type is everywhere in our daily life, completes the network in the user of the available portrait, recommendation system, financial risk assessment, knowledge map, search engine optimization of urban traffic greatly.

Different from the general sense of data mining through the algorithm model (such as the commonly used regression, classification, clustering model) description/prediction, network mining gives a new way: by abstracting the data/problem into a network model, to help us better data analysis/data mining.

Web mining goes one step further with basic descriptive statistics, in a way that reveals a lot of insights that we don’t get with ordinary descriptive analysis.



Many businesses with general data mining methods are not good, but after joining the network model, it can be greatly improved. It can be said that the application stage of network mining is infinitely broad……

For example:

Web page sorting

For example, Google’s PageRank, itself is also based on the construction of a huge web page network model (web pages as nodes, hyperlinks as edges), through the calculation of the center of different pages (weight), to sort the pages, so as to achieve more accurate search and recommendation.

Recommendation system

The basic idea of the traditional collaborative filtering algorithm is to recommend the products favored by users with high similarity to the target users. The addition of network models (such as friend network and commodity network) can largely solve the problems of diversity, cold start and social recommendation, thus improving the recommendation accuracy in some scenarios.

Social Network Analysis

Social networks are naturally suitable for constructing network models for analysis, such as information dissemination prediction, influence analysis, social group discovery, friend recommendation, user portrait, etc., which can be taken out to see individuals individually and find something different from other individuals. To some extent, social network analysis is based on network model analysis.

In fact, network mining is to strengthen the skills of data analysis/mining, with a new perspective, to explore more comprehensive and macro network knowledge, and the relationship between individuals. Network mining can be said to be very hardcore, if you are familiar with network model construction, can obtain:

  • Insights unavailable in other analytical methods to guide actual business decisions;
  • Business models can be constructed through network mining, such as social recommendation and message dissemination model, commodity recommendation system, financial risk control model, etc., which is the real value of data.
  • There are many niche areas where you will earn extra points.

Want to know how many partners learn data analysis, consider giving you a video tutorial for data analysis! You can leave a comment, reply: Python, can receive the full video tutorial data analysis! More words, consider the next period to everyone out!