Data Preprocessing & Feature

What is Data Preprocessing?

Data preprocessing is the process of cleaning, transforming, and preparing raw data so that machine learning algorithms can understand it.

💡 Think of it like cleaning and organizing ingredients before cooking — if the data is messy, the model won’t learn properly.

Steps:

1. Data Cleaning

• Handle missing values (NaN, blanks).
• Remove duplicates.
• Correct inconsistencies (e.g., “Male” vs. “M”).

2. Data Transformation

• Convert categorical values into numbers (One-Hot Encoding, Label Encoding).
• Normalize/standardize numeric values (scale values to the same range).
• Convert text to tokens (for NLP).

3. Data Reduction

• Remove irrelevant columns.
• Dimensionality reduction (PCA, t-SNE).

What are Features?

Features are the measurable properties or characteristics of the data that the model uses to make predictions.

• In a dataset, columns = features (inputs), and one column is usually the target (output).

Example: Predicting house prices

• Features → size, location, number_of_rooms
• Target → price

Feature Engineering

Feature engineering means creating new features or modifying existing ones to improve model performance.

Examples:

• Creating new features: From date_of_birth, create a new feature age. From transaction_amount, create a log(transaction_amount) to handle skewness.

• Feature Selection: Keep only the most relevant features and remove noise (irrelevant columns).

• Encoding categorical features: Example: Convert ["Red", "Blue", "Green"] → [1,0,0], [0,1,0], [0,0,1].

✅ In summary:

• Data preprocessing = cleaning + transforming data.
• Features = inputs that describe the problem.
• Feature engineering = creating/selecting the best features to boost model performance.