Data Preparation

Real-world data is often incomplete or inconsistent and needs to be prepared before modeling. This section covers data cleaning, splitting, scaling and feature preparation techniques.

Introduction

What is Data Preprocessing?Why Data Preprocessing Matters The Data Preprocessing Workflow First Look at a Dataset Building a Preprocessing Pipeline Data Leakage and Preparation Checklist

Data Cleaning

What is Data Cleaning?Common Data Anomalies The Data Cleaning Process Implementation: Inspecting the Titanic Dataset Implementation: Cleaning the Titanic Dataset Validation, Formatting and Best Practices

Missing Values

What are Missing Values?Types of Missing Values Detecting Missing Data Removing Rows with Missing Values Imputation: Mean, Median, Mode and Fill Interpolation and Advanced Imputation Choosing a Method and Its Impact

Outliers

What are Outliers?Detecting Outliers Visually Detecting Outliers: Z-score, IQR and Percentiles Model-Based Outlier Detection Handling Outliers

Splitting Data

Why Split Data?Train/Test and Train/Validation/Test Split Cross-Validation Time-Series and Group-Aware Splits Avoiding Data Leakage and Common Mistakes

Feature Selection

What is Feature Selection?Filter Methods Wrapper Methods Embedded Methods Comparison and Best Practices

Feature Scaling

Why Feature Scaling?Min-Max Scaling (Normalization)Standardization (Z-score Scaling)Robust, Max-Abs, Normalizer and Power Transforms Feature Scaling in Practice

Feature Engineering

What is Feature Engineering?Feature Creation Feature Transformation Encoding Categorical Variables Best Practices and Ideas by Data Type

Feature Extraction

What is Feature Extraction?Principal Component Analysis (PCA)LDA and Other Extraction Techniques Feature Extraction for Text Image and Audio Features, and Best Practices

Imbalanced Data

What is Imbalanced Data?Evaluation Metrics for Imbalanced Data Resampling Techniques Class Weights, Threshold Tuning and Ensembles Choosing an Approach and Common Mistakes