Overview
-
Discover how to solve value estimation problems with machine learning. Learn how to build a value estimation system that can estimate the value of a home.
Syllabus
-
Introduction
- Welcome
- What you should know
- Using the exercise files
- Set up the development environment
- What is machine learning?
- Supervised machine learning for value prediction
- Build a simple home value estimator
- Find the best weights automatically
- Cool uses of value prediction
- Introduction to NumPy, scikit-learn, and pandas
- Think in vectors: How to work with large data sets efficiently
- The basic workflow for training a supervised machine learning model
- Gradient boosting: A versatile machine learning algorithm
- Explore a home value data set
- Standard conventions for naming training data
- Decide how much data you need
- Feature engineering
- Choose the best features for home value prediction
- Use as few features as possible: The curse of dimensionality
- Prepare the features
- Training vs. testing data
- Train the value estimator
- Measure accuracy with mean absolute error
- Overfitting and underfitting
- The brute force solution: Grid search
- Feature selection
- Predict values for new data
- Retrain the classifier with fresh data
- Wrap-up
