Click here to jump straight to notebook.
Let's go through a simple ML application & create our first project. Parallely, we will learn some new concepts & terms.
We will use Iris Dataset that contains 4 features/ measurements (petal length, petal width, sepal length, sepal width
) of 50 samples of 3 species of Iris flower
(Iris setosa, Iris Versicolor, Iris Virginica
) = 50×3 = 150 samples in total. This is one of the best known datasets till date.
For more details on the dataset, visit UCI Machine Learning Repository - IRIS dataset.
For reference, here are pictures of the three flowers species:
Here is picture showing the 4 measurements made from each flower:
Based on these 4 measurements, we can be certain of which species each iris flower belongs to. For a while, we will become hobby botanists trying to classify the flowers into their own species.
Our goal is to build a ML model that learns from the 4 mesurements/ features of the iris flowers whose species is known.
Then, using the learned model, we will predict the species of a new iris flower.
This type of problem is called as Supervised Learning Multi-Class Classification problem
.
Supervised Learning
- We know the correct species (target class) for each iris flower (3 species).
Classification
- We want to predict one of the several classes (the species of iris).Multi Class
- There are 3 types of species we are dealing with, hence the name multi-class.Classes
-
Possible outputs (different species of irises)label
-
For a particular flower, the species it belongs to it is called label.To download the github pages from command line,
curl.exe
from the bin folder and Paste it in C>Program files
& run it once.curl -LJO https://GitHub-URL-to-download/
wget --no-check-certificate --content-disposition https://GitHub-URL-to-download/
« Previous: Introduction to other packages Next: K Nearest Neighbors »