A model is only as good as the data you give it and the way you judge it. This module covers both sides: feature engineering, which turns raw columns into inputs a model can learn from, and model evaluation, which tells you whether the model is actually any good.
You'll start with the techniques that prepare features: handling categorical data with one-hot, label, and target encoding, plus scaling and creating new features from existing ones. Then you'll learn to evaluate honestly using cross-validation, and to read results with the confusion matrix and the ROC curve and AUC.
Once you can build and fairly judge a model, you're ready to explore more powerful approaches in Deep Learning & NLP, the next module in this course.
A beginner-friendly walkthrough of feature engineering, one of the most important steps in a machine learning project, covering how to transform raw data into features a model can learn from.
How to turn categorical columns into numbers a model can use, and the trade-offs between one-hot, label, target, and K-fold target encoding.
Why you can't judge a model on the data it trained on, and how cross-validation lets you compare models and settings in an unbiased way.
How to read a confusion matrix, the table that shows where a classification model gets things right and where it makes mistakes.
How ROC curves and the area under them (AUC) summarize a classifier's performance across all decision thresholds.