Four Real-Life Machine Learning Use Cases: A Databricks guide

Data is the new fuel. The potential for machine learning and deep learning practitioners to make a breakthrough and drive positive outcomes is unprecedented. But how to take advantage of the myriad of data and ML tools now available at our fingertips, and scale model training on big data, for real-life scenarios?

Databricks Unified Analytics Platform is a cloud-service designed to provide you with ready-to-use clusters that can handle all analytics processes in one place, from data preparation to model building and serving, with virtually no limit so that you can scale resources as needed.

This guide walks readers through four practical end-to-end Machine Learning use cases on Databricks:

• A loan risk analysis use case, that covers importing and exploring data in Databricks, executing ETL and the ML pipeline, including model tuning with XGBoost Logistic Regression.
• An advertising analytics and click prediction use case, including collecting and exploring the advertising logs with Spark SQL, using PySpark for feature engineering and using GBTClassifier for model training and predicting the clicks.
• A market basket analysis problem at scale, from ETL to data exploration using Spark SQL, and model training using FT-growth.
• A suspicious behavior identification in videos example, including pre-processing step to create image frames, transfer learning for featurization, and applying logistic regression to identify suspicious images in a video.