Machine Learning is one of the hottest topics in recent times. From self-driving cars to auto-generated codes every exciting technology now uses Machine Learning.
The goal of this blog is to help someone who is trying to get started with machine learning from a technical perspective (coding part) rather than a theoretical perspective. But keep in mind theory / underlying math behind machine learning is as much important as the coding part. Coding works as a tool to implement the theoretical part. I have divided this blog mainly into three parts.
- Basic Understanding of ML algorithms
- Data Processing
- Algorithm implementation
Basic Understanding of ML algorithms
To get started with machine learning you need to have a basic understanding of what machine learning is, the type of problem we can solve using machine learning, some machine learning algorithms, and the underlying intuition behind these algorithms.
For this, I would highly suggest taking Andrew Ng’s Machine learning course on Coursera which is absolutely free. In his course, he teaches algorithms implementation using MATLAB though I would suggest starting with Python as it is highly used in ML and also has a huge community. Also if you prefer learning in Bangla “মেশিন লার্নিং অ্যালগরিদম” by Nafis Nihal is the best book out there.
You should have a basic understanding of how ML works, what are different types of ML techniques, and black box ideas of some well-known machine learning algorithms like KNN, SVM, Random Forest, Linear Regression, etc.
Data Processing
If you are using Python you need to clear the basics of Python programming first. Here is a tutorial if you want to quickly look through the fundamentals of python
One place you can get data to train ML algorithms is Kaggle
The majority of ML data are in CSV format. You can use Pandas library to manipulate CSV data.
Another must-learn Python library for manipulating single or multidimensional arrays/matrices and doing high-level mathematical functions is NumPy
If you want to visualize data manipulation you can use either Matplotlib or Seaborn
Algorithm Implementation
The most used library for implementing Machine Learning algorithms is Scikit Learn
Projects
You can work on the following projects that may help you get a better understanding of Machine Learning
- Weather Prediction
- Movie Recommendation
- Stock Price Predictions
Wrap Up
I utilized the above-mentioned sites to learn about Machine Learning. Feel free to acquire knowledge from any other sources. One technique to learn more quickly is to read the code of others. You can use Kaggle to observe how other programmers write their code.