Machine Learning Overview

Model

is an idealized representation of a system.

- Models don't always match reality, but are still useful

Example of model

- A weather forecast is a model

Machine learning (ML)

is used to generate decision-making and predictivemodels for various purposes

Data collection/analysis

are often preliminary steps in a machine learning project.

Stages of a ML (Machine Learning) Project

1. Problem Definition

2. Data Collection

3. Data Preparation & Preprocessing

4. Building Model

5. Evaluation

6. Deployment

ML (Machine Learning) Algorithms

Models can learn either supervised, unsupervised, or through reinforcement

Supervised Learning

A model is given labelled data and learns underlying relationships between the data and corresponding labels.

Unsupervised Learning

A model is given unlabeled data and identifies patterns in it (ex.clustering). The patterns a model learns may or may not be useful

Reinforcement Learning

An agent learns to make decisions by interacting with an environment. It is used in robotics and other decision-making settings. Ex. roombaML Algorithms

Supervised Learning goal

The goal is to learn a relationship between input variables (features) and an output variable (target)

Supervised Learning examples

regression and classification

Applications:

● Predicting service outage costs (Regression)

● Classifying customer risk (Classification)

Classification

Predicts a categorical outcome.

● Example: Classifying a policyholder as low-risk or high-risk

Regression

Predicts a continuous numerical outcome.

● Example: Predicting the total cost of claims.

Classification vs Regression

What type of machine learning algorithm would be best?

1. Predicting rainfall

2. Predicting gender/sex of a person in an image

3. Predicting an individual's income

4. Predicting the type of animal in an image

5. Diagnosing skin cancer

6. Predicting house prices

1. regression

2. classification

3. regression

4. classification

5. classification

6. regression