In this post, we will be covering the basics of Supervised learning and some of the fundamental ideas you need to know to start learning about Supervised learning.
Supervised learning refers to an algorithm that learns from input to output or x to y. The key characteristic of supervised learning is that you provide the machine learning model with the right answers to learn from. By providing the model with the correct pairs to learn from you can allow the machine learning model to learn and give a reasonably accurate output for new input.
Let's take the example of spam email if you provide a machine learning model with 5000 emails that are correctly labelled as spam or not spam. Then the machine learning model can begin to learn the telling characteristics of what a spam email might contain and will then be able to accurately predict if a new input is a spam or not spam.
Regression
Here is a more tangible example, let's say you want to predict ice cream sales based on the temperature. If you wanted to predict the price for this you could take the data from multiple ice cream carts and map out each plot on a graph. We could then use something called regression to give us a singular line or usually a curve with relative accuracy to inform us of what sales an ice cream vendor can reasonably expect.
Classification
Within supervised learning there is also classification, and we can talk about breast cancer detection as an example. We can classify the tumour as malignant or benign and we are basing this on the size of the tumour.
The way that classification differs from regression is that we are only trying to predict a very small number of outcomes.
To summarise, classification algorithms predict categories, other examples include predicting whether a picture is a cat or a dog.
Conclusion
- Supervised learning maps input x to input y.
- The model learns from the right answers.
- The two main types are regression and classification.
- Regression can have infinite outcomes.
- Classification has a small number of outcomes.
