real-time churn prediction with automl and superset

example_notebooks/Churn-Prediction/inference.ipynb

@@ -0,0 +1,129 @@
+{
+ "cells": [
+  {
+   "attachments": {},
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "#### Import necessary libraries"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import pandas as pd\n",
+    "from autogluon.tabular import TabularPredictor\n",
+    "from sqlalchemy import create_engine"
+   ]
+  },
+  {
+   "attachments": {},
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "#### Specify the path to the test dataset"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "cloud_data_path = \"s3://your-aws-bucket/example\"\n",
+    "test_data_path = f\"{cloud_data_path}/Telco_customer_churn_test.csv\""
+   ]
+  },
+  {
+   "attachments": {},
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "#### Load the trained model"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import s3fs\n",
+    "\n",
+    "def download_model():\n",
+    "    fs = s3fs.S3FileSystem()\n",
+    "\n",
+    "    local_file_path = 'models'\n",
+    "    remote_file_path =f\"s3://your-aws-bucket/demo/{local_file_path}\"\n",
+    "\n",
+    "    # Download the file\n",
+    "    fs.get(remote_file_path, local_file_path)\n",
+    "\n",
+    "    return 'success'"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "download_model()"
+   ]
+  },
+  {
+   "attachments": {},
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "#### Generate predictions"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "predictor = TabularPredictor.load('./models/',require_version_match=False)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "engine = create_engine('postgresql://superset:superset@example/superset')"
+   ]
+  },
+  {
+   "attachments": {},
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "#### Write the predictions to the database"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "test_data_filter.to_sql('churn_predictions', con=engine , if_exists = 'replace')"
+   ]
+  }
+ ],
+ "metadata": {
+  "language_info": {
+   "name": "python"
+  },
+  "orig_nbformat": 4
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}

example_notebooks/Churn-Prediction/pipeline.yaml

@@ -0,0 +1,7 @@
+pipeline:
+  name: "Churn_Prediction_Pipeline"
+  tasks:
+  - name: "train"
+    type: "jupyter notebook"
+    notebook_path: "Churn-Prediction/inference.ipynb"
+    notebook_output_path: "inference_output.ipynb"

example_notebooks/Churn-Prediction/readme.md

@@ -0,0 +1 @@
+Dataset is detailed (here)[https://community.ibm.com/community/user/businessanalytics/blogs/steven-macko/2019/07/11/telco-customer-churn-1113]

example_notebooks/Churn-Prediction/requirements.txt

@@ -0,0 +1,12 @@
+seaborn==0.12.2
+mxnet==1.8.0
+torch==1.13.1+cpu 
+torchvision==0.14.1+cpu 
+autogluon==0.7.0
+imblearn==0.0
+openpyxl==3.1.2
+fastapi==0.70.0
+uvicorn==0.15.0
+boto3==1.17.106
+autogluon.eda==0.7.0
+numpy==1.24.3

example_notebooks/Churn-Prediction/utils.py

@@ -0,0 +1,49 @@
+"""
+helper function
+
+"""
+def download_from_cloud(local_file_name, remote_file_name):
+    """
+    Upload a file to gcs bucket or s3 bucket.
+    """
+    import os
+
+    cloud_name = remote_file_name.split('://')[0]
+    if cloud_name =='gs':
+        import gcsfs
+        fs = gcsfs.GCSFileSystem(project=os.environ['GCP_PROJECT'])
+    elif cloud_name =='s3':
+        import s3fs
+        fs = s3fs.S3FileSystem()
+    else:
+        raise NameError(f'cloud name {cloud_name} unknown')
+    try:    
+        print(f'Downloading from {remote_file_name} to {local_file_name}')
+        fs.get(remote_file_name, local_file_name)
+        print("done downloading!")
+    except Exception as exp:
+        print(f"Download failed: {exp}")
+    return
+
+def upload_to_cloud(local_file_name, remote_file_name):
+    """
+    Upload a file to gcs bucket or s3 bucket.
+    """
+    import os
+    cloud_name = remote_file_name.split('://')[0]
+    if cloud_name =='gs':
+        import gcsfs
+        fs = gcsfs.GCSFileSystem(project=os.environ['GCP_PROJECT'])
+    elif cloud_name =='s3':
+        import s3fs
+        fs = s3fs.S3FileSystem()
+    else:
+        raise NameError(f'cloud name {cloud_name} unknown')
+    try:    
+        print(f'Uploading from {local_file_name} to {remote_file_name}')
+        fs.put(local_file_name, remote_file_name)
+        print("done uploading!")
+    except Exception as exp:
+        print(f"Uploading failed: {exp}")
+
+    return fs