Classification in Supervised Machine Learning

Introduction

Machine learning is changing how we make decisions using data. In ML, supervised learning is one of the most widely used techniques. It involves training a model on labeled data, where the correct answer is already known so that it can make predictions on the new data.

Classification

In Supervised Machine Learning classification is applied where the goal is to predict categories or classes. Given some input data, a model assigns it to one or more categories.

Classification is used to solve problems such as detecting fraud cases in cybersecurity, product recommendation in ecommerce and loan approval in the banking sector.

Types of Classifications

Binary - Only two labels are classified. Example Spam/Not Spam in filtering emails.

Multiclass - More than two classes and each input belongs to a class. Example classifying animals into cats, dogs and so on.

Multilabel - Each input belongs to multiple classes simultaneously. Example a news article can be classified as Political and Financial at the same time.

Popular Classification Algorithms

Logistic Regression - classification algorithm used to predict binary outcomes

Decision Trees - Easy to visualize, works with mixed data. It works by asking a series of yes/no questions (splits) to arrive at a decision

Random Forests - it's a type of bagging ensemble machine learning model built from many Decision Trees.

K-Nearest Neighbors (KNN) - instance-based learning but most commonly, it's used to classify things by asking: "What are the labels of the closest items around this one?"

Classification Steps

1. Data Cleaning - Clean and prepare data so the model can learn.

2. Training the Model - Feed the training data into a classification algorithm. It looks for patterns and decision boundaries that separate the classes.

3. Prediction - The trained model is used to classify new data.

4. Evaluation - Evaluate data using metrics such as accuracy, Recall, Precision and F1 score.

My Take on Classification
I was really fascinated to learn that binary is used in classification, The first and most fundamental language of computers is binary and now learning that it is essential in ML makes it interesting.

In Conclusion Classification is about teaching machines to learn from labeled data so they can make accurate predictions and informed decisions in the real world. It turns data into decisions.