1.
Introduction
2.
Dependencies
3.
Setup the model template
4.
Run the model template
5.
How to create your own model algorithm
6.
Implementing the model algorithm
6.1.
Adding or removing dependency libraries
6.2.
Setting up the editor
6.2.1.
Vim
6.2.2.
VSCode
6.3.
Train a model
6.3.1.
Sanitize input
6.3.2.
Feature engineering
6.3.2.1.
Label encoding
6.3.2.2.
Time component encoding
6.3.2.3.
Remove datetime columns
6.3.2.4.
Handle holdout dataframe
6.3.2.5.
One-hot encoding
6.3.3.
Extract unique labels
6.3.4.
Calculate cross validation scores
6.3.5.
Train the machine learning model
6.3.6.
Calculate a single set of scores
6.3.7.
Save a model
6.3.7.1.
Saving scikit-learn stuff
6.3.7.2.
Creating ModelInfo instance
6.3.7.3.
Encrypting & saving the ModelInfo
6.3.8.
Return the results
6.3.9.
Piecing it together
6.4.
Predicting with a trained model
6.4.1.
Load a model
6.4.2.
Sanitize input
6.4.3.
Apply feature engineering
6.4.4.
Predict
6.4.5.
Format the results
6.4.6.
Piecing it together
7.
Testing the model algorithm
7.1.
Running unit tests
7.2.
Debugging the model algorithm
7.2.1.
Debugging the deployed model algorithm
7.3.
Using the landing page
8.
Publishing the model algorithm
9.
Modifying the model algorithm
9.1.
Migrating the ModelInfo
9.2.
Bumping the version
10.
Appendix A: Helper functions and classes in exodusutils
10.1.
Enums
10.2.
Constants
10.3.
Schemas
10.3.1.
Requests
10.3.2.
Scores
10.4.
Exceptions
10.5.
Frames
10.6.
Feature engineering
10.7.
Frame manipulation
10.8.
Miscellaneous
10.9.
How to update exodusutils
11.
Appendix B: How cross validation scores are calculated
12.
Appendix C: How to change MongoDB settings
13.
Appendix D: How to view the API docs
Light (default)
Rust
Coal
Navy
Ayu
Exodus Model Template Documentation
Feature engineering
one_hot_encoding
label_encoding
time_component_encoding
See previous chapters for explanation.