I have got this question many times by people who wants to start with Machine learning but don't know a roadmap so I thought of writing an article on this. When you are a beginner, things might look blur but everything is achievable if you put efforts. Below I’m going to tell you how to kick off studying Machine Learning.
Why Machine Learning?
We are living in era where everything is depended on machines/computers, now getting work done by machines is not sufficient we want machines to learn and think like humans so they can do work without being dependent on humans.
Machine Learning enables computers to learn and do job without taking much directions from humans. Machine Learning algorithms learn from real world and take decision itself without any human interaction which makes this more useful and future of machines.
Steps for beginners to get started:
Step 1: Choose a programming language:
If you are not from programming background then you must learn a programming language. You can choose Python/R which has large communities and many libraries available already. Even learning python is very easy which is good news for you.
If you have good programming background you can continue with your current programming language, like Java, C# also having communities, libraries for Machine Learning and they are growing it. You no need to worry about programming language since it’s about studying algorithms.
If you are an experienced programmer it will be very easy to start with programming language like python. Python is great for machine learning and R is great fit for data science. You can easily google which tools are required to kick off.
Start with Python:
Start with R:
Start with C#, F#:
Start with Java:
Step 2: Get basic knowledge of Mathematics
Having knowledge of basic math is required and once you go in depth good knowledge of mathematics is required. Initially if you have basic knowledge of 2D, Calculus, Statistics, Probability, Algebra it is good to start without any prior preparation.
If you come from non-mathematics background first you should learn basic math. Not depth knowledge is required but basics are must to understand and apply mathematical formulas.
Step 3: Follow a Book/Blog/Tutor
Be focused on what you are doing, take a book/blog which gives you initial direction. Do not try to learn everything from multiple sources. Choose one thing and complete it first, once you get basics done you can look for other sources. Following sources, where you can start:
Step 4: Practice
“Practice makes a man perfect” slogan fits best here. Practicing problems, visualizing data, building datasets, finding useful information in data, training algorithms, correcting your code and every single effort you put brings you more understanding.
Step 5: Use Libraries
Once your basics are clear you can start using libraries. There are a lot of libraries out there, may be for face recognition, plotting data points, clustering etc. Do not start directly with libraries as beginner, first learn algorithms and as soon you feel confident you can start with libraries.
Step 6: Projects/POCs
This step is important. Once you are done with learning you should get into building projects, a project is always different than studying. A project will give you more understanding on where to apply which algorithms, real-time scenarios, business need etc.
Above steps will help you getting started. For any other help/queries you may write in comments.