The purpose of this article is to explain how to utilize the models tab.
Overview
The Models (Beta) menu allows you to train and save multivariate models to the Petro DB. Once a model is built, it can be utilized for predictions at any point in the future.
When getting started, you will initially see two tabs. The first tab is called "Loaded Models." In this tab, you can see all of the historical models previously created with their respective creation and update dates. From this menu, you can click "Load" and bring in the historical models in which you are interested, or you can refresh to see the current status of the models loaded.
The next tab is called "New Models." In this tab, you need to enter a name of your choice for the new model, and a new tab will pop-up by the name "Edit: <Your Model Name>.
Requirements
- RAI Product Version - 3.1
- PetroDatabase (Schema Version - 3.0)
New/Edit Model
When editing a new or existing model, you will be presented with the menu below:
Field | Description |
Name | Name of the model |
Description | Description of the model |
Trainer | Selected trainer of the model. Selections include but are not limited to Random Forest and Multiple Linear Regression. Some Trainers will have additional options (Example: Random Forest will require the user to input the number of trees used for the training.) |
Test Set/Test Set Size (%) | Method by which to create and size the testing set |
Table/Columns | Table and Columns used for the model training |
After adding inputs and outputs, you can Save and Train the model. A few things to note:
- Saving will save the model to PetroDatabase and make it accessible to others.
- The model's training results and errors can be viewed in the PetroSuite.
- Training will create a predicted column in the data table in the following format: <Predicted Column>_<Name of Model>
Predict Model
After training, the model is ready to be used. In the Loaded Models tab, find the model of interest and select predict.