Machine learning is an interesting field . Machine learning is the study of algorithms and it is seen as a part of artificial intelligence and we can solve many problems through machine learning such as speech recognition , translation , computer vision as well as conducting groundbreaking research for the betterment of people’s lives and improvement of the healthcare system etc . Machine learning algorithms build a model based on sample data , known as training data , in order to make predictions or decisions without being heavily programmed to do so . To put it in simple words machine learning is where you teach the computer based on data (input) to predict new output .
Are you stuck on ml even after learning it for so much time . This is when you know that you lack a lot of theoretical skills and the depth of the field . Astonishingly even after having so many courses , guides, books many people end up abandoning machine learning or they stop learning ml due to various reasons . Or even after learning they cannot fully apply their skills to their dream ml project . I personally got to experience this while i was learning ml for 2 months but i skipped many topics thinking they weren’t of that importance but when i got to really implement my skills on a project then i knew about my lack of machine learning . After a particular stage you can’t understand the complex algorithms and functions which will make your learning process way harder .
Below i am going to share you how to learn ml my way . This means i will guide you about the process of learning ml i am also going to share some links of some courses . First we will cover all the topics required for basic ml then I will upload blogs about intermediate and advanced ml soon.
1) Math
Math is the most important topic in ml . If you skip it then you will not understand ml . Math is the foundation of ml . It provides you with an overlook over the functions , algorithms , how models work etc . The question is much how much math ? As for beginner level i will highly recommend learning the three big topics they are as follows Linear Algebra , Calculus , Statistics
Linear Algebra
Linear Algebra is very important topic in ml it covers the field of neural networks and it is all about basic operations and algorithms . It will get you started about the basics of computers and how a computer works etc . In linear algebra you should learn all the Pre-calculus stuff . The khan academy pre calculus would be enough because it covers all the matrix , functions , graphs etc .
Calculus
A lot of ml is dependent on calculus . Calculus is an important topic in ml . It covers the working of models, functions mainly such as loss function , cost function etc . Calculus will let you know about the functions mainly such minima , maxima . You will learn how to implement models and algorithms in a correct manner and what algorithms to implement .
Statistics
Statistics is a very vast subject and it is also very important in ml . But need not worry you don’t need to know every topic in statistics but only the basics to get you started . Summaries , Similarities , Probabilities and its distribution functions etc . are enough to get you started . Statistics is mostly non as non neural network machine learning .
This math will not only open your doors to one field but many other fields such as competitive programming , artificial intelligence , deep learning etc
2) Programming
You have learnt all the math and are now ready to build some machine learning models and some groundbreaking research . But sadly Programming too is hard and it will take you time to master programming . You have to spend much of your time in collecting dataset , cleaning it , editing it , then do some research on it to find out which model to use then feature selection , preprocessing , validating and predicting it . Below i have written Python because i find it easy and it also has many various other applications such as in django etc . There is language also for ml it is known as R . But it is your wish both Python and R are similar but R is more ml centric .
Python
Python is the best beginner programming language for machine learning because it includes many libraries and packages which are made especially for ml . Some packages which you need to master are numpy , pandas , regex (for NLP) , sklearn. There are many other packages also but these packages are enough for you to get started . Intermediate python as in functions , loops , slicing , indexing , editing the datasets , running different models etc will be enough as of now .
Working with data
Data is a key part of ml . You have to collect it through web scraping or some datasets will be publicly available . Next you have to fill in missing values , rename columns and rename values for simplicity , then you have to work with data visualization . Data visualization is where you will visualize your data with the help of several packages such as matplotlib and seaborn . During this part you will see many similarities and you will get to know many key factors and correlation with different factors .
3)Implementing
Finally you’ve learnt all the math and programming . Now comes the last and the hard part implementing the skills in real world problems and projects . Many people get confused in between what packages to choose and how will they differ these packages . Go with your heart read the documentations of these packages and see which one do you like the most .
Following are the 7 steps of machine learning
Collecting Data
Data Preparation
Choosing model
Training
Evaluating
Parameter tuning
Predicting
Now what can we do with these skills there are many problems as of now to get started on but i will first recommend doing the challenges on kaggle . The best problem to get started is on the house price prediction challenge it gives you an good overview of how machine learning works to solve real world problems . In the challenge you are give a dataset for which you should seperate the data in train, test and then train the data and then evaluate it and then you should predict the price .
Following are the links which i recommend for you to learn ml :
Math
These links are of the khan academy one but if you want to you can also check out 3blue1brown Youtube channel and his series on maths .
Programming
Hackerrank is not for learning but for practicing python there are beginner level questions and advanced ones also . But for learning python I highly recommend you to learn them from youtube there isn’t one specific favorite video . Most of them are nice if you want a recommendation then I would recommend freecodecamp video . And to practice python Iwould recommend using vscode for starting .
Thanks for reading this article . If you have any suggestions please feel free to comment .
Implementing
Implementing is the best part . The best website is kaggle for implementing there are many competitions , notebooks , datasets , etc . You can also look winning solutions for past compeitions .
With this I wish you all the best for your machine learning journey .
I am also still a beginner in machine learning . Feel free to comment and give suggestions and tell me if I have missed anything . Any suggestion would be highly appreciated .