Introduction to Machine Learning
Hey!
I will be posting a series of posts that will introduce you to the basics of machine learning. Also, since I myself started with ML just 2 months back, I will be encountering all the possible doubts and errors that you might get, both in theory and practice. I’ll be writing about linear and logistic regression, gradient descent, Naive Bayes (Probabilistic approach), neural networks, SVM, decision trees and feature selection (PCA).
So let’s dive into it.
“ Machine learning (ML) is the scientific study of algorithms and statistical models that computer systems use to perform a specific task without using explicit instructions, relying on patterns and inference instead. ” — Wikipedia
In simple words, ML is the science of making the system learn and work without explicitly defining the task for the machine. The system learns by taking feedback from history i.e. historical data.
There are 3 branches of ML:
Supervised Learning — the target of every data point is available and the system learns through feedback from every data point. Used for regression, classification etc.
Unsupervised Learning — the target of any data point is not given. The machine learns by observing patterns in the given data. Used for pattern recognition, data clustering etc.
Reinforcement Learning — the machine learns by attempting different approaches. Every approach has a reward or loss. The system learns by trying to gain the highest reward. Used to automate systems.
The steps that are followed for all problems are:
Step 1- Collecting and Preparing Data — You can get a lot data sets from online sources such as kaggle. Also, you can use library embedded data sets from libraries like sklearn or you can even generate artificial data.
Step 2- Choosing and Training Model — We will be looking at different machine learning algorithms in the future posts, and I will also compare all of them as for you to decide which model can be used depending on the type data.
Step 3- Evaluation of Model — After training our model on training data, we can also test our model and evaluate it to find how accurate the model is.
I’ll be mainly focusing on classification and regression. Classification refers to the process through which the system can predict the class label of a given data point. While regression predicts a continuous value.
So this was an introduction to ML, I’ll be explaining linear and logistic regression in the next post from theory to practical. Check out the next story for more :)
It is not a necessity but do take a look at how some of the python libraries such as numpy and matplotlib.pyplot work.
