Introduction to machine learning

Big Data: Technological and social changes have transformed the amount and types of data that are produced, and our capacity to process them. We have:

• Volume of data
• Variety of data
• Velocity to process it

For data to be useful, we need to process it efficiently to extract information that is useful and make decisions.

Machine learning: Type of data analysis that aims to use as much data information as possible with as little structure as needed.

Classification by type of ML task:

- Supervised learning
- Unsupervised learning
- Reinforcement learning.

Examples of ML models:

- Linear and non-linear regression
- Classification and regression trees
- Neural networks

Training

A model is trained to perform a given ML task by following the steps below:

1. Data acquisition and feature selection
2. Training (optimisation).
3. Validation and hyper-parameter optimisation (go to 2. it needed).
4. Testing.

This requires a good management of data to avoid over-fitting