adds mlflow example

doc/_toc.yml

@@ -35,6 +35,7 @@ parts:
       - file: apps/dash
       - file: apps/django
       - file: apps/jupyterlab
+      - file: apps/mlflow
       - file: apps/quarto
       - file: apps/docker
       - file: apps/chalk-it

doc/apps/jupyterlab.md

@@ -2,15 +2,15 @@
 
 
 ```{important}
-If you're on a free account, always back up your work because your app can be terminated due to inactivity. If you need JupyterLab deployed with production settings, contact us [[email protected]](mailto:contact.ploomber.io)
+If you're on a free account, always back up your work because your app can be terminated if inactive. If you need JupyterLab deployed with production settings, contact us [[email protected]](mailto:contact.ploomber.io)
 ```
 
 You can deploy JupyterLab to Ploomber Cloud and use it as a development environment.
 
-
 To do so, download the files from the
-[example](https://github.com/ploomber/doc/tree/main/examples/docker/jupyterlab),
-and follow the instructions for [Docker deployments.](docker.md)
+[example](https://github.com/ploomber/doc/tree/main/examples/docker/jupyterlab), create a `.zip` file and deploy it using the Docker option:
+
+![](../static/docker.png)
 
 Once the deployment finishes, go to `WEBSERVICE LOGS` and copy the token:
 

doc/apps/mlflow.md

@@ -0,0 +1,92 @@
+# MLflow
+
+```{important}
+If you're on a free account, always back up your work because your app can be terminated if inactive. If you need MLflow deployed with production settings, contact us [[email protected]](mailto:contact.ploomber.io)
+```
+
+To deploy MLflow download the Dockerfile from the
+[example](https://github.com/ploomber/doc/tree/main/examples/docker/mlflow), create a `.zip` file and deploy it using the Docker option:
+
+![](../static/docker.png)
+
+Once the deployment finishes, open the app by clicking on `VIEW APPLICATION`, if all
+went well you'll be prompted for a user and a password:
+
+Use: admin
+Password: password
+
+## Changing the default password
+
+
+For safety, let's change the credentials. Install the `requests` package:
+
+```sh
+pip install requests
+```
+
+And run the following in a Python session:
+
+```python
+import requests
+from getpass import getpass
+
+host = input("Enter your {id}.ploomberapp.io URL: ")
+password_new = getpass("Enter new password:")
+
+response = requests.patch(
+    f"{host}/api/2.0/mlflow/users/update-password",
+    auth=("admin", "password"),
+    json={"username": "admin", "password": password_new},
+)
+
+response.raise_for_status()
+
+print("Your new password is:", password_new)
+```
+
+You'll be prompted for your MLflow host (e.g., `something.ploomberapp.io`), you can
+get this from the Ploomber UI:
+
+![](../static/docker/mlflow/host.png)
+
+Then, enter the new password.
+
+To confirm that the password was changed, open the MLflow dashboard again and enter the new password.
+
+
+## Tracking experiments
+
+To start tracking experiments, install mlflow locally:
+
+```sh
+pip install mlflow
+```
+
+Set your credentials as environment variables:
+
+```sh
+export MLFLOW_TRACKING_USERNAME=admin
+export MLFLOW_TRACKING_PASSWORD=yourpassword
+```
+
+And run the following (replace the first line with your host!):
+
+```python
+import mlflow
+
+mlflow.set_tracking_uri(uri="https://someid.ploomberapp.io")
+
+mlflow.set_experiment("first-experiment")
+
+with mlflow.start_run():
+    mlflow.log_params({"a": 21, "b": 21})
+    mlflow.log_metric("accuracy", 0.91)
+```
+
+If all goes well, you'll see:
+
+```
+2024/03/12 19:44:39 INFO mlflow.tracking.fluent: Experiment with name 'first-experiment' does not exist. Creating a new experiment.
+```
+
+And the experiment will be visible from MLflow's dashboard.

examples/docker/mlflow/Dockerfile

@@ -0,0 +1,7 @@
+FROM python:3.11-slim
+
+RUN pip install mlflow
+
+
+
+ENTRYPOINT ["mlflow", "server", "--host", "0.0.0.0", "--port", "80", "--app-name", "basic-auth"]