Introduction to Machine Learning

# CHAPTER 21

Introduction to Machine Learning #

1. Chapter Introduction #

2. What is Machine Learning? #

Traditional Programming relies on hardcoded rules: *Data + Rules = Output* (e.g., If Age > 18: output "Adult")

Machine Learning flips this. You give the computer the data and the historical answers, and it *learns the rules itself*: *Data + Historical Answers = Learned Rules (A Model)*

Once the model is trained, you can give it new data, and it will use its learned rules to predict the output.

3. Supervised vs. Unsupervised Learning #

There are two main branches of Machine Learning:

1. Supervised Learning: The data has a known "answer" or "label" (This is 90% of ML jobs).

• Regression: Predicting a continuous number. (e.g., Predicting House Prices).

• Classification: Predicting a category. (e.g., Is this email Spam or Not Spam?).

2. Unsupervised Learning: The data has NO labels. The algorithm must find hidden structures on its own.

• Clustering: Grouping similar data together. (e.g., Customer Segmentation based on shopping habits, without knowing who they are).

4. The Standard ML Workflow #

Every Machine Learning project follows this exact pipeline:

1. 1. Prepare Data: Clean NaNs, scale numbers, encode text to numbers.

1. 2. Split Data: Separate data into a Training Set (to learn) and a Testing Set (to evaluate).

1. 3. Choose Algorithm: Pick a Scikit-Learn model (e.g., Linear Regression).

1. 4. Fit (Train): Pass the Training Data to the model so it can learn.

1. 5. Predict: Ask the model to guess the answers for the Testing Data.

1. 6. Evaluate: Compare the model's guesses to the real answers to score its accuracy.

5. Introduction to Scikit-Learn #

Scikit-Learn (sklearn) is the premier machine learning library for Python. It contains hundreds of algorithms, all accessible via a unified, simple syntax.

Installation:

!pip install scikit-learn

6. The Scikit-Learn API Design (The 3-Step Process) #

Scikit-Learn is famous for its elegant, consistent design. Every single algorithm works using these same three steps:

# 1. IMPORT the algorithm you want to use
from sklearn.linear_model import LinearRegression

# 2. INITIALIZE the model (Create an instance of it)
model = LinearRegression()

# 3. FIT the model (Train it by giving it Data (X) and Answers (y))
# model.fit(X_train, y_train)

# 4. PREDICT (Ask it to guess on new data)
# predictions = model.predict(X_test)

7. Terminology: X and y #

In Machine Learning notation:

• X (Capital X): The Features. The data you are using to make the prediction (e.g., Square Footage, Number of Bedrooms, Zip Code). This is usually a 2D Pandas DataFrame.

• y (Lowercase y): The Target. The answer you are trying to predict (e.g., House Price). This is usually a 1D Pandas Series.

8. Common Mistakes #

• Thinking ML is magic: ML algorithms are just complex statistical equations. If you feed them terrible, messy data, they will learn terrible, messy rules. (Garbage In, Garbage Out).

• Using Classification for Numbers: Trying to use a Logistic Regression (Classification) model to predict a House Price (Regression). You must understand your Target variable before choosing an algorithm.

9. MCQs #

10. Interview Questions #

• Q: Explain the difference between Supervised and Unsupervised Machine Learning. Give a business use-case for both.

• Q: What is the difference between Classification and Regression tasks?

11. Summary #

12. Next Chapter Recommendation #