class: center, middle, inverse, title-slide .title[ # MLflow ] .author[ ### Justin Post ] --- layout: false class: title-slide-section-red, middle # MLflow Justin Post --- layout: true <div class="my-footer"><img src="data:image/png;base64,#img/logo.png" style="height: 60px;"/></div> --- # Documenting Model Fitting Process - Fitting many models, we should make a log of the results/information! <img src="data:image/png;base64,#img/mlflow_diagram.jpg" width="700px" style="display: block; margin: auto;" /> --- # Documenting Model Fitting Process - Fitting many models, we should make a log of the results/information! - [`MLflow` is a great package](https://mlflow.org/docs/latest/index.html) that does this and more - Not included in our jupyterhub, must install it first - Open a new Terminal and run `pip install mlflow` (and perhaps some other things) <img src="data:image/png;base64,#img/mlflow_install.jpg" width="500px" style="display: block; margin: auto;" /> --- # `MLflow` Basics - Let's go through their tutorial to understand the process! --- # Recap - `MLflow` gives us a nice way to document the model fitting process - Has a nice GUI but not easy to use on our jupyterhub - Can easily convert results into a pandas data frame and inspect! ```python import mlflow mlflow.search_runs(experiment_ids=["0"]) ```