🧠 Machine Learning Foundations – Statistics, Probability & Data Preparation

This repository contains foundational learning materials to help you understand data before training machine learning models.
If we don’t understand data first, our models may become biased, unstable, or misleading — so statistics is the first step.

---

📌 What is Statistics?

Statistics is the science of collecting, summarizing, analyzing, and interpreting data.

Example:
If we have students' exam scores:

• Mean → Average score
• Median → Middle score
• Standard Deviation → How spread out the scores are

Statistics helps us find meaning in data.

---

🤖 Why Statistics is Important for Machine Learning

Machine learning models learn patterns from data.
If the data has:

• Outliers
• Skewed distribution
• Wrong scaling
• Missing values

Then the model will learn wrong patterns.

Statistics helps us:

• Understand center & spread
• Detect outliers
• Identify skewness and long tails
• Scale features & encode categories
• Evaluate model performance correctly

---

📚 Course Modules

Module 01 — Descriptive Statistics & Distributions

Topic	Purpose
Mean, Median, Mode	Measure center of data
Variance & Standard Deviation	Measure spread
Percentiles & Quartiles	Understand rank within dataset
IQR (Interquartile Range)	Outlier detection
Z-Score	Standardization
Distribution Shapes	Symmetric vs Skewed vs Long-tail

Use Median + IQR when data is skewed or contains outliers.
Use Mean + SD when data is symmetrical.

---

Module 02 — Probability Basics for ML

• Events, outcomes, sample space
• Conditional probability & independence
• Bayes’ Theorem (foundation for Naive Bayes)
• Sensitivity, specificity, false positives/negatives
• Class imbalance problems

---

Module 2.5 — Practice Worksheets

• Compute mean, median, SD, IQR, fences
• Z-score & outlier detection (manual + Python)

---

Module 03 — Data Quality, Scaling & Encoding

Concept	Why It Matters
Missing Data Types (MCAR/MAR/MNAR)	Correct imputation
Min-Max, Standard & Robust Scaling	Prevents unfair feature influence
One-Hot & Ordinal Encoding	Proper handling of categorical data
Distance Metrics	Used in KNN, Clustering, Embeddings
Covariance & Correlation	Feature relationship understanding
PCA (concept intro)	Dimensionality reduction

---

Module 3.5 — Hands-on Worksheets

• Bayes rule problems
• Confusion matrix: Precision, Recall, F1-Score

Module 04 — Quiz / Review

Conceptual wrap-up before applying ML algorithms.

---

🎯 Week 1 Goal

Build intuition, not memorize formulas.

We learn to:

• Summarize data
• Detect outliers
• Understand real-world distributions
• Prepare data for ML models to be accurate, robust & explainable

---

📂 Folder Structure