Data Prep Basics: Rows, Columns, and Targets
- Predictive AI
- H2O Driverless AI
This lesson introduces the foundational concepts behind data preparation for classical machine learning, starting with why poor data quality leads to poor models and slower projects. It explains the classical tabular structure of rows and columns, the difference between supervised learning, which has a target variable, and unsupervised learning, which does not, and walks through classification and regression using a fraud detection and a product pricing example. It also introduces the unit of analysis, the entity each row represents, which shapes how a dataset should be built before it goes into Driverless AI.
Key takeaways
- Why poor data quality leads directly to poor models, bad conclusions, and slower project timelines
- How supervised learning differs from unsupervised learning based on whether a target variable exists
- What separates a classification problem from a regression problem based on the type of target variable
- How the unit of analysis, like a transaction or a product, determines what each row in a dataset represents
Start owning your AI