You are on page 1of 15

Machine Learning 1:

Supervised and Unsupervised Learning

Samir Tata, PhD


samir@horizon-university.tn
Outline
● Who am I
● Introduction to Python and useful tools and libraries
● Supervised Learning
○ Introduction to Supervised Learning
○ Model Validation
○ Building Regression From Scratch
○ Classification
● Unsupervised Learning
○ Clustering

WWW.HORIZON-TECH.TN
HORIZON SCHOOL OF DIGITAL TECHNOLOGIES
Introduction to Supervised ML

3
Machine Learning
Supervised Unsupervised Reinforcement

Business Examples Prediction Anomaly detection Reward Systems


Fraud detection Consumer behavior Recommendation systems
Recommendation Image Recognition
Techniques Regression Clustering Value/Policy optimization
Classification Deep Learning
Deep Learning
Regression Classification Clustering

Linear Regression SVM K-Means

GLM Discriminant Analysis K-Medoids


SVR Naïve Bayes Fuzzy C-Means
GPR Nearest Neighbor Hierarchical
Decision Trees Gaussian Mixture
Neural Networks Neural Networks
Ensemble Methods Hidden Markov Models

4
Supervised Learning
• The data comes with the correct answer
• The model created is used to predict the new unknown values
• Example
• Train a model for predicting car prices based on car attributes using historical
sales data. That model can then predict the optimal/potential price for new
cars that haven’t been sold before

5
Linear Regression

6
Linear Regression

7
Linear Regression

8
Linear Regression

9
Linear Regression

10
Mean Absolute Error

11
Mean Square Error

12
Mean Absolute Percentage Error

13
Regression
• Linear
• Fit a line to the data observations (y = ax + b)
• Use this line to predict observation values
• Multivariate
• More that one variable influences the one you’re interested in
• Example: predicting the price of a care based on its attributes (body style, brand,
etc.)
• Polynomial
• y = ax2 + bx + c
• y = ax3 + bx2 + cx + d
• Nonlinear

14
EXPAND YOUR HORIZON

WWW.HORIZON-TECH.TN
HORIZON SCHOOL OF DIGITAL TECHNOLOGIES

You might also like