Python：pmml Introduction to format file 、 install 、 Usage method ( utilize python Turn the machine learning model into Java frequently-used pmml Format file ) It's a detailed introduction
PMML（Predictive Model Markup Language） It is mainly used as the carrier of analyzing model training examples , Up to PMML 4.2 edition , Supported models include ： Association rules 、 Baseline model 、 Decision tree 、 clustering 、 Return to 、KNN、 neural network 、 Bayes 、 Scoreboard 、 Sequence 、 Text 、 The time series 、 Rule sets and SVM. PMML Use a unified specification for validation , Use XSD Do lexical verification , Use XSLT Do grammar validation , In the construction and analysis, we should follow the unified PMML Standard operation .
PMML It's a de facto standard language , For rendering data mining models . Predictive analysis model And data mining models A term for a mathematical model , These models use statistical techniques to understand patterns hidden in large amounts of historical data . The prediction analysis model uses the knowledge acquired in the process of finalization to predict whether there is a known pattern in the new data .PMML Allows you to easily share predictive analysis models between different applications . therefore , You can shape a model in a system , stay PMML To express it in , And then move it to another system , In the system, the above model is used to predict the possibility of machine failure .
Simply speaking ,PMML It's training models on a platform , And then it's packaged as PMML file , Then we can use the trained model directly in another platform .PMML It's the product of data mining groups , The group is a committee led by suppliers , It's made up of a variety of commercial and open source analysis companies . therefore , Most of today's leading data mining tools can be exported or imported PMML. As a developed 10 Years of mature standards ,PMML It can present statistical techniques for understanding models from data （ Such as artificial neural network and decision tree ）, It can also present the preprocessing of the original input data and the post-processing of the model output .