Classification Algorithms (Decision trees, SVM, Logistic regreession)

What are Classification Algorithms?

Classification is a method in Data Science where we predict which category or group a data point belongs to. For example, predicting whether an email is spam or not, or whether a patient has a disease

1. Decision Trees

How it works:

1. Works like a question-answer game.
2. It asks questions like “Is age > 30?” and splits data into branches.
3. The last leaf gives the final class (answer).

Use cases:

1. Predict if a customer will leave
2. Medical diagnosis
3. Loan/credit risk check

Pros (good points):

1. Very easy to understand
2. Works with numbers and text
3. Less data cleaning needed

Cons (bad points):

1. Can overfit (memorize training data too much)
2. Small changes in data can make a very different tree

2.Support Vector Machines (SVM)

How it works:

1. Imagine you have red and blue dots on paper.
2. SVM draws a line (boundary) to separate them with maximum distance between groups.
3. This line is called a hyperplane.

Use cases:

1. Image recognition
2. Text categorization (like spam detection)
3. Bioinformatics (like classifying proteins)

Pros (good points):

1. Works well when there are many features
2. Memory efficient (uses less space)

Cons (bad points):

1. Not good for very large datasets
2. Not good with noisy or overlapping data

3.Logistic Regression

How it works:

1. Gives a probability between 0 and 1 using an S-shaped curve (sigmoid)
2. If probability is > 0.5 → class A, else class B

Use cases:

1. Spam detection
2. Disease prediction
3. Will a customer buy a product or not

Pros (good points):

1. Simple, fast, and easy
2. Easy to explain

Cons (bad points):

1. Works best when data is linear
2. Sensitive to outliers (unusual data points)