Machine Learning

Overview#

Machine learning is a way to predict decisions based on different sets of data. Using data, you can decide how to structure queries and then predict information based off of those queries.

ML relies on understanding and identifying patterns in data. Capturing patterns in data is called fitting or training a model.

The initial data used to create the model, is called the training data.

Decision Trees, the basics#

Models are built commonly off of Decision Trees. Decision Trees are a way of sorting data that makes logical sense based upon a pattern.

Decision trees can be built with as few as two decisions.

Parts of an Oreo Cookie ├── Cookie └── Filling

Not all decisions have one factor#

As with life, (and Oreos for that matter) not all things are black and white (or cookie and filling). While the parts of an Oreo Cookie may in fact be a binary, in which there are only two outcomes, not all things are.

Let's look at this decision tree below and explore how it might relate to say... airline seat prices.

First off, with an airline seating chart you have a few different factors at play. Did you splurge and get that First Class experience, or are you riding economy?

Is it a window seat? Aisle seat? or did you get stuck in the middle?

What about emergency exit rows? Front or back of the plane? Next to the bathrooms? These all are things that can impact the price of the decision tree.

Airline Seat Price Tree#

Airline Seat Prices ├── First Class │ ├── Aisle Seat │ └── Window Seat ├── Business Class │ ├── Aisle Seat │ ├── Window Seat │ └── Middle Seat └── Middle Class ├── Aisle Seat ├── Window Seat └── Middle Seat

Using the tree built, you can predict the price of your seat based on the different characteristcs provided. The end price will be what it comes to at the bottom of the tree, or the leaf.

Decision Trees aren't the only model used in ML#

Decision trees are not the only model used in machine learning, there are other models commonly used.

These include: