Decide how much data you need

Contents

• Introduction Introduction

  • Welcome

    42s
  • What you should know

    21s
  • Using the exercise files

    31s
  • Set up the development environment

    2m 21s
• 1. What Is Machine Learning and Value Prediction? 1. What Is Machine Learning and Value Prediction?

  • What is machine learning?

    3m 11s
  • Supervised machine learning for value prediction

    2m 45s
  • Build a simple home value estimator

    2m 37s
  • Find the best weights automatically

    4m 7s
  • Cool uses of value prediction

    2m 6s
• 2. An Overview of Building a Machine Learning System 2. An Overview of Building a Machine Learning System

  • Introduction to NumPy, scikit-learn, and pandas

    1m 22s
  • Think in vectors: How to work with large data sets efficiently

    2m 58s
  • The basic workflow for training a supervised machine learning model

    2m 22s
  • Gradient boosting: A versatile machine learning algorithm

    3m 55s
• 3. Training Data 3. Training Data

  • Explore a home value data set

    2m 56s
  • Standard conventions for naming training data

    53s
  • Decide how much data you need

    2m 4s
• 4. Features 4. Features

  • Feature engineering

    4m 11s
  • Choose the best features for home value prediction

    3m 19s
  • Use as few features as possible: The curse of dimensionality

    1m 50s
• 5. Coding Our System 5. Coding Our System

  • Prepare the features

    1m 48s
  • Training vs. testing data

    1m 3s
  • Train the value estimator

    2m 51s
  • Measure accuracy with mean absolute error

    1m 31s
• 6. Improving Our System 6. Improving Our System

  • Overfitting and underfitting

    2m 44s
  • The brute force solution: Grid search

    2m 46s
  • Feature selection

    2m 26s
• 7. Using the Estimator in a Real-World Program 7. Using the Estimator in a Real-World Program

  • Predict values for new data

    2m 39s
  • Retrain the classifier with fresh data

    1m 48s
• Conclusion Conclusion

  • Wrap-up

    48s