Photo by Fotis Fotopoulos on Unsplash
 

I enrolled into an undergraduate computer science program and decided to major in data science. I spent over $25K in tuition fees over the span of three years, only to graduate and realize that I wasn’t equipped with the skills necessary to land a job in the field.

I barely knew how to code, and was unclear about the most basic machine learning concepts.

I took some time out to try and learn data science myself — with the help of YouTube videos, online courses, and tutorials. I realized that all of this knowledge was publicly available on the Internet and could be accessed for free. 

It came as a surprise that even Ivy League universities started making many of their courses accessible to students worldwide, for little to no charge. This meant that people like me could learn these skills from some of the best institutions in the world, instead of spending thousands of dollars on a subpar degree program.

In this article, I will provide you with a data science roadmap I created using only freely available MIT online courses.

 

Step 1: Learn to code

 
I highly recommend learning a programming language before going deep into the math and theory behind data science models. Once you learn to code, you will be able to work with real-world datasets and get a feel of how predictive algorithms function. 

MIT Open Courseware offers a beginner-friendly Python program for beginners, called Introduction to Computer Science and Programming. 

This course is designed to help people with no prior coding experience to write programs to tackle useful problems.

 

Step 2: Statistics

 
Statistics is at the core of every data science workflow — it is required when building a predictive model, analyzing trends in large amounts of data, or selecting useful features to feed into your model.

MIT Open Courseware offers a beginner-friendly course called Introduction to Probability and Statistics. After taking this course, you will learn the basic principles of statistical inference and probability. Some concepts covered include conditional probability, Bayes theorem, covariance, central limit theorem, resampling, and linear regression.

This course will also walk you through statistical analysis using the R programming language, which is useful as it adds on to your tool stack as a data scientist.

Another useful program offered by MIT for free is called Statistical Thinking and Data Analysis. This is another elementary course in the subject that will take you through different data analysis techniques in Excel, R, and Matlab.

You will learn about data collection, analysis, different types of sampling distributions, statistical inference, linear regression, multiple linear regression, and nonparametric statistical methods.

 

Step 3: Foundational Math Skills

 
Calculus and linear algebra are two other branches of math that are used in the field of machine learning. Taking a course or two in these subjects will give you a different perspective of how predictive models function, and the working behind the underlying algorithm.

To learn calculus, you can take Single Variable Calculus offered by MIT for free, followed by Multivariable Calculus.

Then, you can take this Linear Algebra class by Prof. Gilbert Strang to get a strong grasp of the subject.

All of the above courses are offered by MIT Open Courseware, and are paired with lecture notes, problem sets, exam questions, and solutions.

 

Step 4: Machine Learning

 
Finally, you can use the knowledge gained in the courses above to take MIT’s Introduction to Machine Learning course. This program will walk you through the implementation of predictive models in Python.

The core focus of this course is in supervised and reinforcement learning problems, and you will be taught concepts such as generalization and how overfitting can be mitigated. Apart from just working with structured datasets, you will also learn to process image and sequential data.

MIT’s machine learning program cites three pre-requisites — Python, linear algebra, and calculus, which is why it is advisable to take the courses above before starting this one.

 

Are These Courses Beginner-Friendly?

 
Even if you have no prior knowledge of programming, statistics, or mathematics, you can take all the courses listed above.

MIT has designed these programs to take you through the subject from scratch. However, unlike many MOOCs out there, the pace does build up pretty quickly and the courses cover a large depth of information.

Due to this, it is advisable to do all the exercises that come with the lectures and work through all the reading material provided.

 
 
Natassha Selvaraj is a self-taught data scientist with a passion for writing. You can connect with her on LinkedIn.