Classification algorithm is one of the most commonly used data mining methods, and its core idea is to find out the * * * same characteristics of target data items, and classify data items into different categories according to classification rules. Clustering algorithm divides a group of data into several categories according to similarity and difference, which makes the similarity between the same type of data as large as possible and the similarity between different types of data as small as possible. The purpose of classification and clustering is to classify data items, but there are significant differences between them. Classification is supervised learning, which means that these categories are known. By training and learning the classified data, we can find the characteristics of these different categories, and then classify the unclassified data. But clustering is unsupervised learning, and there is no need to train and learn data. Common classification algorithms include decision tree classification algorithm and Bayesian classification algorithm. Clustering algorithms include systematic clustering and K-means clustering.
2. Regression analysis
Regression analysis is a statistical analysis method to determine the quantitative relationship between two or more variables, and its main research problems include the trend characteristics of data series, the prediction of data series and the correlation between data. According to the number of independent variables in the model, regression algorithms can be divided into univariate regression analysis and multivariate regression analysis; According to the relationship between independent variables and dependent variables, it can be divided into linear regression analysis and nonlinear regression analysis.
3. Neural network
Neural network algorithm is a network system developed on the basis of modern neurobiological research to simulate the information processing mechanism of human brain. It not only has general computing ability, but also has the ability to think, learn and remember knowledge. It is a learning algorithm based on tutor, which can simulate the input and output of complex systems and has very strong nonlinear mapping ability. The mining process based on neural network includes four stages: data preparation, rule extraction, rule application and prediction evaluation. In data mining, neural network algorithm is often used for prediction.
4. Correlation analysis
Association analysis is to find the association, correlation or causal structure between item sets or object sets in transaction data, relational data or other information carriers, that is, to describe the relationship rules between different data items in the database. For example, if one data item changes and the other changes, there may be some correlation between the two data items. Correlation analysis is a very useful data mining model, which can help enterprises to output many useful product portfolio recommendations and preferential promotion portfolios, find potential customers and truly realize data mining. 4 The application of marketing big data mining in precision marketing can be divided into two categories, offline application and online application. Among them, offline applications mainly carry out data mining based on customer portraits, and carry out targeted marketing activities for different purposes, including potential customer mining, lost customer retention, and refined marketing media. Online applications based on real-time data mining results, accurate advertising push and marketing, including DMP, DSP and programmatic buying.