Adding experiment to MLflow_and_NeuralProphet.ipynb (#1499)

* Update MLflow_and_NeuralProphet.ipynb Adding example with MLflow Autologging, dataset, requirements & environment metadata with model signature inference * Updating commentary Updating commentary and fixing test related issues (e.g. docs build test). * Updating documented code Changed if local loop start. * Unexpected indentation Resolving unexpected indentation issue --------- Co-authored-by: Oskar Triebe <ourownstory@users.noreply.github.com>
ourownstory · Feb 13, 2024 · f140e1e · f140e1e
1 parent 305de42
commit f140e1e
Showing 1 changed file with 94 additions and 9 deletions.
diff --git a/docs/source/how-to-guides/feature-guides/MLflow_and_NeuralProphet.ipynb b/docs/source/how-to-guides/feature-guides/MLflow_and_NeuralProphet.ipynb
@@ -32,8 +32,7 @@
    "outputs": [],
    "source": [
     "# for this tutorial, we need to install MLflow.\n",
-    "#!pip install mlflow\n",
-    "# import mlflow\n",
+    "!pip install mlflow\n",
     "\n",
     "# Start a MLflow tracking-server on your local machine\n",
     "# !mlflow server --host 127.0.0.1 --port 8080\n",
@@ -42,7 +41,7 @@
     "    # uninstall preinstalled packages from Colab to avoid conflicts\n",
     "    !pip uninstall -y torch notebook notebook_shim tensorflow tensorflow-datasets prophet torchaudio torchdata torchtext torchvision\n",
     "    !pip install git+https://github.com/ourownstory/neural_prophet.git  # may take a while\n",
-    "\n",
+    "    \n",
     "# much faster using the following code, but may not have the latest upgrades/bugfixes\n",
     "# pip install neuralprophet"
    ]
@@ -58,6 +57,7 @@
    "source": [
     "import pandas as pd\n",
     "from neuralprophet import NeuralProphet, set_log_level, save\n",
+    "import mlflow\n",
     "import time\n",
     "\n",
     "set_log_level(\"ERROR\")\n",
@@ -177,6 +177,7 @@
     "# Start a new MLflow run\n",
     "if local:\n",
     "    with mlflow.start_run():\n",
+    "\n",
     "        # Create a new MLflow experiment\n",
     "        mlflow.set_experiment(\"NP-MLflow Quickstart_v1\")\n",
     "\n",
@@ -185,8 +186,8 @@
     "\n",
     "        # Define NeuralProphet hyperparameters\n",
     "        params = {\n",
-    "            \"n_lags\": 5,\n",
-    "            \"n_forecasts\": 3,\n",
+    "        \"n_lags\": 5,\n",
+    "        \"n_forecasts\": 3,\n",
     "        }\n",
     "\n",
     "        # Log Hyperparameters\n",
@@ -199,12 +200,12 @@
     "        end = time.time()\n",
     "\n",
     "        # Log training duration\n",
-    "        mlflow.log_metric(\"duration\", end - start)\n",
+    "        mlflow.log_metric('duration', end-start)\n",
     "\n",
     "        # Log training metrics\n",
-    "        mlflow.log_metric(\"MAE_train\", value=list(metrics_train[\"MAE\"])[-1])\n",
-    "        mlflow.log_metric(\"RMSE_train\", value=list(metrics_train[\"RMSE\"])[-1])\n",
-    "        mlflow.log_metric(\"Loss_train\", value=list(metrics_train[\"Loss\"])[-1])\n",
+    "        mlflow.log_metric('MAE_train', value=list(metrics_train['MAE'])[-1])\n",
+    "        mlflow.log_metric('RMSE_train', value=list(metrics_train['RMSE'])[-1])\n",
+    "        mlflow.log_metric('Loss_train', value=list(metrics_train['Loss'])[-1])\n",
     "\n",
     "        # save model\n",
     "        model_path = \"np-model.np\"\n",
@@ -234,6 +235,90 @@
     "In order to see the results of our run, we can navigate to the MLflow UI. Since we have already started the Tracking Server at http://localhost:8080, we can simply navigate to that URL in our browser and observe our experiments. \n",
     "If we click on the respective experiments we can see a list of all runs associated with the experiment. Clicking on the run will take us to the run page, where the details of what we’ve logged will be shown."
    ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Example with MLflow Autologging, dataset, requirements & environment metadata with model signature inference, another MLflow Experiment with NeuralProphet"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# MLflow setup\n",
+    "# Run this command with environment activated: mlflow ui --port xxxx (e.g. 5000, 5001, 5002)\n",
+    "# Copy and paste url from command line to web browser\n",
+    "\n",
+    "import mlflow\n",
+    "import torchmetrics\n",
+    "from mlflow.data.pandas_dataset import PandasDataset\n",
+    "\n",
+    "if local:\n",
+    "\n",
+    "    mlflow.pytorch.autolog(log_every_n_epoch=1, log_every_n_step=None,\n",
+    "    log_models=True, log_datasets=True, disable=False,\n",
+    "    exclusive=False, disable_for_unsupported_versions=False,\n",
+    "    silent=False, registered_model_name=None, extra_tags=None\n",
+    "    )\n",
+    "\n",
+    "    import mlflow.pytorch\n",
+    "    from mlflow.client import MlflowClient\n",
+    "\n",
+    "    model_name = \"NeuralProphet\"\n",
+    "\n",
+    "\n",
+    "    with mlflow.start_run() as run:\n",
+    "\n",
+    "        dataset: PandasDataset = mlflow.data.from_pandas(df, source=\"AirPassengersDataset\")\n",
+    "\n",
+    "        # Log the dataset to the MLflow Run. Specify the \"training\" context to indicate that the\n",
+    "        # dataset is used for model training\n",
+    "        mlflow.log_input(dataset, context=\"training\")\n",
+    "\n",
+    "        mlflow.log_param(\"model_type\", \"NeuralProphet\")\n",
+    "        mlflow.log_param(\"n_lags\", 8)\n",
+    "        mlflow.log_param(\"ar_layers\", [8, 8, 8, 8])\n",
+    "        mlflow.log_param(\"accelerator\", \"gpu\")\n",
+    "\n",
+    "        #To Train \n",
+    "        # Import the NeuralProphet class\n",
+    "        from neuralprophet import NeuralProphet, set_log_level\n",
+    "\n",
+    "        # Disable logging messages unless there is an error\n",
+    "        set_log_level(\"ERROR\")\n",
+    "\n",
+    "        # Create a NeuralProphet model with default parameters\n",
+    "        m = NeuralProphet(n_lags=8,  \n",
+    "              ar_layers=[8, 8, 8, 8],\n",
+    "              trainer_config = {\"accelerator\":\"gpu\"})\n",
+    "\n",
+    "        # Use static plotly in notebooks\n",
+    "        m.set_plotting_backend(\"plotly-resampler\")\n",
+    "\n",
+    "        # Fit the model on the dataset\n",
+    "        metrics = m.fit(df)\n",
+    "\n",
+    "        df_future = m.make_future_dataframe(df, n_historic_predictions=48, periods=12)\n",
+    "\n",
+    "        # Predict the future\n",
+    "        forecast = m.predict(df_future)\n",
+    "\n",
+    "    mlflow.metrics.mae()\n",
+    "\n",
+    "    # Save conda environment used to run the model\n",
+    "    mlflow.pytorch.get_default_conda_env()\n",
+    "\n",
+    "    # Save pip requirements\n",
+    "    mlflow.pytorch.get_default_pip_requirements()\n",
+    "\n",
+    "    # Registering model\n",
+    "    model_uri = f\"runs:/{run.info.run_id}/NeuralProphet_test\"\n",
+    "    mlflow.register_model(model_uri=model_uri, name=model_name)"
+   ]
   }
  ],
  "metadata": {