Explore the Models panel to learn more about the mechanics and performance of prediction models you build. Your project will only feature Models after you build a prediction model using the Models → Train & Predict analysis type.

‍

‍

“As the statistician George E. P. Box wrote, "All models are wrong, but some models are useful." What he meant by that is that all models are simplifications of the universe, as they must necessarily be."

- Nate Silver, The Signal and the Noise

‍

If you are looking to understand some of the technical terms used in Models - check out this article in our Technical Docs.

‍

Creating Models

In order to use the Models panel, you need to build a project using the Models → Train & Predict analysis type. Choose a dataset to work with, select the model's target and factors and execute your project using the setup wizard.

‍

‍

How to Create a Model?

1. Start from the 'Datasets' panel of your Graphext workspace.
2. Select a dataset to start working with it.
3. Pick 'Models' as your type of analysis from the left sidebar.
4. Select the option to 'Train and predict' a model.
5. Inside the 'Clusters and Network Creation' card that appears, start by adding a target.
6. To add a target, select the variable from the list on the right side of the 'Clusters and Network Creation' card.
7. Click 'Send Here' on the first box under the 'Target' column.
8. Next, add a factor by selecting a variable from the list on the right side of the 'Clusters and Network Creation' card.
9. Click 'Send Here' on the first box under the 'Factors' column.
10. Add more factors by repeating steps 8 - 9.
11. Review the other cards in the project setup sidebar before selecting 'Continue'.
12. Done ... Your model will make predictions on your target variable once you have executed your project.

‍

Opening Model Tabs

Once you've built a model in Graphext, you'll be able to navigate to the Models panel using your project's top menu. Click on a tab to open it.

‍

‍

How to Open Model Tabs?

1. Start from inside a project where you used Models → Train & Predict as your analysis type.
2. Select Models from your project's top menu.
3. Click on any of the three tabs - General - Training - Result - to reveal the information inside the tab.
4. Done ... Examine your model's accuracy, parameters and associated notes.

‍

Checking the Library Behind a Model

Models you build in Graphext are built on top of trusted open-source technologies that are widely used in many data science capacities. You can check which open-source library is behind a model you deployed inside of the General tab belonging to your Models panel.

‍

How to Check the Library Behind a Model?

1. Start from inside a project where you used Models → Train & Predict as your analysis type.
2. Select Models from your project's top menu.
3. Click on the General tab to open it.
4. Find the value next to the Library key.
5. Clicking on the value will take you to the documentation behind the open-source library.
6. Done ... Explore the documentation to learn more about the mechanics behind your model.

‍

Reusing a Model

You can reuse a model you have trained on new datasets with identical column names and variable types. When you build a model, it will be stored in your Graphext workspace making it accessible in other projects that you create within the same team.

To reuse a model, you need to set the name of the model you want to use inside of the advanced editor. You can find which models are available to you by inspecting the autofill menu that appears after you start typing inside of the "model" parameter field.

‍

How to Reuse a Model?

1. Start from the 'Datasets' panel of your Graphext workspace.
2. Select a dataset to start working with it. This must have identical column names and variable types to the dataset you used to train your model.
3. Pick 'Models' as your type of analysis from the left sidebar.
4. Select the option to 'Train and predict' a model.
5. Inside the 'Clusters and Network Creation' card that appears, add your variables as factors and targets. These should be the same as the factors and targets you used to train your model.
6. Select 'Open Advanced Editor.
7. Inside the advanced editor, find the 'train' step.
8. Directly underneath, start typing 'predict'.
9. Click the 'predict' step from the autofill menu.
10. The first parameter currently reads 'data'. Copy and paste the first parameter from the 'train' step in place of the first parameter of the 'predict' step. Both should now include a reference to your targets and factors.
11. Inside the second parameter of the 'predict' step, enter the name of your model.
12. Rename the output of your model so that it reads 'ds.gx_prediction' rather than 'ds.predicted'.
13. Delete the 'train' step.
14. Done ... Click continue to make predictions on your new dataset using your old model.

‍