MLModel Pattern, a.k.a. Machine Learning Model Pattern
To create the machine learning model's definition and data provider from a Transaction object, to be used with GeneXusAI to train a custom model that can be later used to make predictions on new data.
Many problems can be solved by using a machine learning model.
The MLModel Pattern supports models for Regression and Classification problems.
Regression problems try to predict a continuous value from input data. The canonical example is the case of predicting the selling price of a house, given information such as the number of bedrooms, the total surface, the neighborhood it is in, etc.
Classification problems try to predict a category from the input data. The canonical example is to predict wether an email is spam or not.
The pattern is generated from a Transaction object, and by default it tries to identify which of the Transaction's attributes may be inputs and output of the model.
Given that information, the pattern then generates two Data Provider objects;
- The first Data Provider returns a GeneXusAI.Custom.Definition object defining the model structure
- The second one returns a collection GeneXusAI.Custom.Data and supports paging as expected by the GeneXusAI module.
You can find the pattern sources in GitHub.
To install the pattern follow the instructions to build it in the GitHub repository's readme file.