Beginners Learning Path for Machine Learning

#machinelearning #datascience #tensorflow #python

Made your mind towards machine learning but are confused so much that where to get started. I faced the same confusion that what should be a good start? Should I learn Python, or go for R? Mathematics was always a scary part for me and I was always worried that from where should I learn math? I was also worried that how should I get a strong basis for Machine Learning. Anyways you should be congratulated that at least you have made your mind.

In this article, I will tell you some basic steps and courses you should take before starting practical machine learning.

Is Mathematics 1st step for Machine Learning?

1st Step for Machine Learning ( Basic Programming)

The very 1st step is that you should learn programming and most probably Python. If you have never written a single line of code I will recommend you is CS50 by Harvard University, the best course for beginners in programming, it will teach you from scratch till all the C Language and a lot of Python and JavaScript with basics of SQL and JSON. This very exciting course is free at edx.org and you can find it here. Read an article about cs50 here

If you want to take start directly from Python then I’ll recommend you to take MIT’s Introduction to Computer Science with Python available on edx for free. Notice that high school algebra is a prereq for this course because they will make you solve a lot of mathematical real-world problems.

If you feel that you computational logic is not good and you are not a good problem solver in programming, I’ll recommend you this course by Microsoft which is all related to Computational Thinking available on edx all for free. This course is really cool and teaches a lot of computational logic and critical thinking.

If you are good enough with basics of programming but have no knowledge of object-oriented programming, then I recommend you to learn Object-Oriented Programming. Though it is not used a lot in machine learning(basic level) but certainly it helps a lot. I will suggest you this course which teaches you the basics of oop and algorithms in python. You can find this course for free at edx.org.

Next step is familiarity with data structures and algorithms. A good programmer must know some basic algorithms like a linked list, binary trees, etc. This course Microsoft will teach you

Algorithmic Analysis
Sorting and Searching of Algorithms
Data Structures: Linked lists, Stacks, Queues

This course is available for free at edx.org.

If you want to go deep in Data Structures and Algorithms, UC San Diego’s specialization in is a masterpiece. It has 6 courses and will take you from 0 to hero level in data structures and algorithms. You can click on each course and audit its material for free but if you want the certificate, its price is 50$/month, depends on you that how early you can complete it. Find this specialization in Coursera.org or click here.

2nd Step For Machine Learning

Yes, you guessed it right, the most dangerous part for most of the beginners, MATHEMATICS.

But don’t worry , it isn’t that difficult as we think of it and if you think you have good basics of high school math(vectors, matrices, calculus, probability, and stats) you can simply just take a refresher course but if you think you are not good enough then the best place for anyone to learn math is KHAN ACADEMY. The amount of content is so rich and helpful at khan academy and everything is available for free. You can find all linear algebra, probability and stats, and multivariable calculus courses there.

Another good course is Math for machine learning specialization from coursera by Imperial College London. It is a great course which teaches basics and revises concepts but does not dive too deep. Exercises and quizzes are quite challenging. It has 3 courses on it.

Mathematics for Machine Learning: Linear Algebra
Mathematics for Machine Learning: Multivariate Calculus
Mathematics for Machine Learning: PCA

MIT has a great course on Statistics which teaches

Construct estimators using the method of moments and maximum likelihood, and decide how to choose between them
Quantify uncertainty using confidence intervals and hypothesis testing
Choose between different models using the goodness of fit test
Make prediction using linear, nonlinear and generalized linear models
Perform dimension reduction using principal component analysis (PCA)

Linear Regression
Polynomial Regression
Logistic Regression
Multi-class Classification
Neural Network
Support Vector Machine (SVM)
K-means Clustering
Primary Component Analysis (PCA)
Anomaly Detection
Recommender System

The only problem (according to some people) is that it is taught in octave/Matlab but for me its no big deal, because it will clear all the basics and deep concepts of machine learning that implementing them in any other language, is not going to be any sort of problem plus Matlab is an additional skill. This course is available for free on Stanford’s official website and at coursera.

This course is a must-go for everyone like it is soo good that it can’t be explained.

After this course now you need to implement all the things you learned in python and go advanced beyond what you learned so specialization by deeplearning.ai on coursera by our own Andrew Ng, Kian Katanfrosh, and Younes Bensouda is also a must take. This specialization has 5 courses which are

Neural Networks and Deep Learning
Improving Deep Neural Networks: Hyper-parameter tuning, Regularization, and Optimization
Structuring Machine Learning Projects
Convolutional Neural Networks
Sequence Models

Main skills you are going to gain are

TensorFlow
Convolutional Neural Network
Artificial Neural Network
Deep Learning

Okay so now you have good grip on TensorFlow and machine learning and you might not like TensorFlow or you just want to test the famous contestant of TensorFlow which is PyTorch(by Facebook) so why not start this free deep learning course by Udacity offered by Facebook in collaboration with Amazon Web Services and part of Udacity’s famous deep learning Nano Degree worth 1400$.

Of course, this is just a start and there is a lot more to learn and discover but if you do these things hopefully you will get an idea of where you stand and what to do next.
Some Advises and Recommendations:

Here are some recommendations.

Start listening to “Machine Learning Guide” Podcast by OC Devel right now. This is going to give you a complete overview of machine learning, what is in it and the journey of an android and web applications developer shifting towards the path of machine learning. Available on Apple Podcasts/iTunes, Google Podcasts and his website.
Get “Hands-on Machine Learning with Scikit-Learn and TensorFlow: Concepts, Tools, and Techniques to Build Intelligent Systems” or its new version releasing on October 2019 named as “Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow: Concepts, Tools, and Techniques to Build Intelligent Systems”. Author of this book is Aurélien Géron. 1st version of this book is #1 in Artificial Intelligence (Books), #1 in Natural Language Processing (Books) & #1 in Computer Neural Networks. 2nd version is not released yet but available for pre-order. This book is very extensive and recommended. Order it now from Amazon.
Don’t forget to read this article. It has a lot of useful information.
Practice a lot on Kaggle(You will know about it once you start ML).
Be dedicated and believe that you can do it. All you need is consistency.