In supervised learning, we will be having the set of sample inputs and outputs. Our job is to create an algorithm with given data based on different parameters (that influence the generation of outputs).
Regression ProblemsIn the problems which fall under this category, the algorithm will be predicting an approximate value for the output.
For example, consider a problem where we need to predict share price of stock in future based on its past records. The possible parameters could be - company's import, export data, changes in consumer markets etc.
Classification ProblemsIn this category, based on different parameters we will classify data to discrete valued outputs. So there won't be a prediction of output values, Instead will map to an output group.
For example, consider a simple spam filter which categories e-mail as spam or not based on the presence of few words.
If we are considering multiple parameters, the plot will be like this.
On next - Unsupervised learning
Continuation - Linear Regression