Data Preprocessing in Machine Learning: Working with Numerical & Categorical Data

2026-05-31 · Source: Data Science on Medium · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics · Depth: Novice, medium

Summary

This guide introduces essential data preprocessing techniques for machine learning, emphasizing its critical role before model training. It details methods for handling numerical data, including filling missing values using mean or median, applying feature scaling through normalization (0-1 range) or standardization (mean/spread adjustment), and detecting/treating outliers by removal, capping, or transformation. For categorical data, the guide explains label encoding (categories to numbers), one-hot encoding (creating binary columns), and ordinal encoding (preserving order), alongside strategies for rare or unknown categories. The content also covers general data cleaning, feature transformation, real-world applications in healthcare and banking, and common preprocessing mistakes, highlighting how proper preparation improves model accuracy and efficiency.

Key takeaway

For data scientists preparing datasets for model training, prioritizing robust data preprocessing is crucial. You should meticulously handle missing values, scale numerical features appropriately, and encode categorical data using methods like one-hot or ordinal encoding. Ignoring these steps can lead to unreliable predictions and inefficient model learning. Always check for outliers and duplicates to ensure your model learns from the most accurate and balanced data possible.

Key insights

Data preprocessing is fundamental for machine learning model accuracy, ensuring clean, transformed data for effective learning.

Principles

• Clean data improves model accuracy.
• Models learn better from scaled numerical data.
• Categorical data needs numerical conversion.

Method

The article describes a general workflow: identify data types (numerical/categorical), handle missing values, scale numerical features, encode categorical features, detect/treat outliers, and clean for duplicates/errors.

In practice

• Fill missing numerical values with mean/median.
• Apply one-hot encoding for nominal categories.
• Group rare categories to simplify datasets.

Topics

• Data Preprocessing
• Numerical Data Handling
• Categorical Data Encoding
• Feature Scaling
• Outlier Detection
• Machine Learning Workflows

Best for: AI Student, Data Scientist, Machine Learning Engineer

Related on AIssential

• From Messy to Clean: 8 Python Tricks for Effortless Data Preprocessing — Efficient Python tricks streamline data preprocessing, addressing common challenges with concise code.
• 5 categorical encodings, 5 wrong answers: one-hot, label, target, ordered target, and what CatBoost actually does — Categorical features are ubiquitous in tabular ML projects.
• From Raw Data to Intelligent Models: Feature Engineering, Backpropagation, and the Power of… — Data preparation, including feature engineering and visualization, is paramount for machine learning model success.
• Data Leakage in Machine Learning: Why You Must Split Before Preprocessing — Preprocessing steps that learn from data must occur after splitting to prevent data leakage and ensure valid model evaluation.
• Building Robust Credit Scoring Models (Part 3) — Proper data splitting and preprocessing are crucial for building generalizable and stable credit scoring models.
• Why Data Cleaning Takes Longer Than the Model In AI — High-quality data preparation is more critical for machine learning success than complex algorithms.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Data Science on Medium.