Introduction to Feature Engineering
Chapter 1: The Role of Features in Machine Learning
Revisiting the Machine Learning Workflow
What Constitutes a Feature?
Impact of Feature Quality on Model Performance
Common Data Types and Their Challenges
Overview of Feature Engineering Tasks
Quiz for Chapter 1
Chapter 2: Handling Missing Data
Identifying Missing Values
Mechanisms of Missing Data (MCAR, MAR, MNAR)
Simple Imputation Strategies: Mean, Median, Mode
Creating Missing Value Indicators
Multivariate Imputation: KNN Imputer
Multivariate Imputation: Iterative Imputer
Comparing Imputation Methods
Hands-on Practical: Imputing Missing Data
Quiz for Chapter 2
Chapter 3: Encoding Categorical Features
Challenges with Categorical Data
Nominal vs. Ordinal Categories
One-Hot Encoding for Nominal Features
Ordinal Encoding for Ordered Features
Handling High Cardinality Features
Target Encoding (Mean Encoding)
Binary Encoding
Hashing Encoder
Comparing Encoding Methods
Hands-on Practical: Applying Encoding Techniques
Quiz for Chapter 3
Chapter 4: Feature Scaling and Transformation
The Need for Feature Scaling
Standardization (Z-score Scaling)
Normalization (Min-Max Scaling)
Scaling for Outliers
Log Transformation for Skewed Data
Box-Cox Transformation
Yeo-Johnson Transformation
Quantile Transformation
Choosing the Right Scaling/Transformation Method
Hands-on Practical: Scaling and Transforming Features
Quiz for Chapter 4
Chapter 5: Feature Creation
Motivation for Creating New Features
Interaction Features
Polynomial Features
Feature Creation from Date/Time Data
Binning Numerical Features
Domain-Specific Feature Engineering
Automated Feature Creation (Introduction)
Hands-on Practical: Engineering New Features
Quiz for Chapter 5
Chapter 6: Feature Selection
Importance of Feature Selection
Filter Methods Overview
Filter Methods: Variance Threshold
Filter Methods: Univariate Statistical Tests (ANOVA F-value, Chi-Squared)
Filter Methods: Correlation Analysis
Wrapper Methods Overview
Wrapper Methods: Recursive Feature Elimination (RFE)
Wrapper Methods: Sequential Feature Selection (SFS)
Embedded Methods Overview
Embedded Methods: Regularization (Lasso L1)
Embedded Methods: Tree-Based Feature Importance
Hands-on Practical: Selecting Features
Quiz for Chapter 6
Log Transformation for Skewed Data
Was this section helpful?
- Applied Predictive Modeling, Max Kuhn and Kjell Johnson, 2013 (Springer) - This book provides comprehensive coverage of data preprocessing techniques, including various transformations like log transformation, explaining their role in enhancing model performance.
- Python for Data Analysis: Data Wrangling with Pandas, NumPy, and IPython, Wes McKinney, 2022 (O'Reilly Media) - A practical guide to data manipulation using Python's Pandas and NumPy libraries, directly relevant for implementing numerical transformations such as the log transformation.
- numpy.log1p, NumPy Developers, 2023 - Official documentation for NumPy's
log1pfunction, detailing its mathematical definition, usage, and why it's suitable for non-negative data including zeros. - An Analysis of Transformations, George E. P. Box and David R. Cox, 1964 Journal of the Royal Statistical Society. Series B (Methodological), Vol. 26 (Oxford University Press) DOI: 10.1111/j.2517-6161.1964.tb00553.x - This foundational paper introduces the Box-Cox transformation, which generalizes power transformations including the log transformation, offering a statistical framework for improving data normality.