01 - Machine Learning Overview

Machine Learning

A sub-field of artificial intelligence where rules are automatically “learned” from data, rather than explicitly programmed

Supervised Learning

The model tries to predict a label (also called target)

Supervised Learning

You can measure if the model is good by comparing its predictions to the “correct answer”

Supervised Learning

There is a “correct answer” to every example

Regression

Supervised Learning: If the model is trying to predict something numerical / continuous

Classification

Supervised Learning: If the model is trying to predict something categorical / discrete

Dataset

Supervised Learning Terms: Collection of data instances

Instance

Supervised Learning Terms: A single row / object in the dataset

Label

Supervised Learning Terms: The target variable that we want to predict

Classes

Supervised Learning Terms: List of possible values for the target variable (classification only). In this case: {iris-setosa, iris-versicolor, iris-virginica}

Features

Supervised Learning Terms: The variables that will be considered in making a prediction.

Feature Vector

Features Intuition: Numerical representation of a single instance

Input

Supervised Learning Framework: Feature vector of instance to be predicted

ML Model

Supervised Learning Framework: The machine learning model (sometimes called hypothesis function) is a function that maps a given input to a corresponding prediction

Prediction

Supervised Learning Framework: Numerical value for regression, a class label for classification

Unseen data

Train-Test Split: Testing the model must use _____, otherwise performance results may be biased

Random

Train-Test Split: Split must be _____

Artificial Intelligence

A field of research in computer science that focuses on making computers exhibit intelligent behavior

Traditional AI

Humans give the rules based on how they see the problem/data

Machine Learning

Model/techniques use statistics, calculus, linear algebra

Unsupervised Learning

There is no label. No “correct answer” for each example in the data

Unsupervised Learning

The goal is to simply find patterns from the data

Collect data

Basic Supervised ML Pipeline

1. _____

2. Preprocess data (exploratory data analysis, cleaning, etc.)

3. Identify features and label.

4. Split data into training set and test set.

5. Build and fine-tune model from the training set.

6. Run the test set on the model to measure its performance.

7. Iterate as needed

Preprocess data

Basic Supervised ML Pipeline

1. Collect data

2. _____

3. Identify features and label.

4. Split data into training set and test set.

5. Build and fine-tune model from the training set.

6. Run the test set on the model to measure its performance.

7. Iterate as needed

Identify features and label

Basic Supervised ML Pipeline

1. Collect data

2. Preprocess data (exploratory data analysis, cleaning, etc.)

3. _____

4. Split data into training set and test set.

5. Build and fine-tune model from the training set.

6. Run the test set on the model to measure its performance.

7. Iterate as needed

Split data

Basic Supervised ML Pipeline

1. Collect data

2. Preprocess data (exploratory data analysis, cleaning, etc.)

3. Identify features and label

4. _____ into training set and test set.

5. Build and fine-tune model from the training set.

6. Run the test set on the model to measure its performance.

7. Iterate as needed

Build and fine tune model

Basic Supervised ML Pipeline

1. Collect data

2. Preprocess data (exploratory data analysis, cleaning, etc.)

3. Identify features and label

4. Split data into training set and test set.

5. _____ from the training set.

6. Run the test set on the model to measure its performance.

7. Iterate as needed

Run the test set

Basic Supervised ML Pipeline

1. Collect data

2. Preprocess data (exploratory data analysis, cleaning, etc.)

3. Identify features and label

4. Split data into training set and test set.

5. Build and fine-tune model from the training set.

6. _____ on the model to measure its performance.

7. Iterate as needed

Iterate as needed

Basic Supervised ML Pipeline

1. Collect data

2. Preprocess data (exploratory data analysis, cleaning, etc.)

3. Identify features and label

4. Split data into training set and test set.

5. Build and fine-tune model from the training set.

6. Run the test set on the model to measure its performance.

7. _____