How to Data Sciecne Without a Degree

Read the original article here.

INTRODUCTION

Hey there.

I want to show you how to become a Data Scientist without a degree (or for free). Ironically, I do have a degree — one that was even made for Data Science (Master’s in Analytics from Northwestern).

But to give you a little background, I used to be an accountant at Deloitte. Isn’t that crazy? I was far from data science or anything technical. I had to learn a lot of things online on my own after work and even during my Master’s program to catch up to my peers’ level as I came from a non-technical background. Having gone through the experience myself, I can tell you that a degree is very helpful, but not necessary.

Because I have been on both sides of getting a degree and learning things online, I think I can give you a unique perspective. Getting a Master’s in data science is a sure and fast way to get into the field, but luckily you don’t have to if you don’t want to spend $60–90k on tuition. It will require a lot of your self-discipline though.

If a friend asks me how to get into data science, this post would be for them. I hope you find my advice valuable and relevant as I have gone through the process myself and found these resources useful. Before we get into the details, let’s find out what data science is about.

WHAT DO YOU DO AS A DATA SCIENTIST?

From my experience working as a data scientist at a few companies like GoDaddy, HERE, and GoGo, data scientists solve problems by applying machine learning on big data.

Some examples include:

Predicting customers’ probability to cancel a subscription
Identifying data anomalies
Computing ad-hoc analysis on gigabytes or terabytes of data
Clustering customers into meaningful groups
Text analytics to find topics in customer chat transcripts
Calculating revenue projections

As a data scientist, you get thrown a lot of different types of problems. To be competent, you need to have a strong foundation in math, statistics, and programming. You need to know when to use certain techniques and algorithms depending on the problem and the data. At the end, you often need to present the results and techniques to executives and less-technical audiences.

Also, as a data scientist, you need to continue to learn and adapt. Because the field is changing rapidly, it is important to stay up-to-date and learn new techniques. Even today, I spend a lot of time studying.

WHAT IT TAKES TO BECOME A DATA SCIENTIST (FOR FREE)

Does data scientist work sound exciting to you? Great. This is a good time to be alive to learn for free. I tried to focus on free or cheap options because who doesn’t like free things? It just takes your commitment and perseverance. I will describe this process in three phases.

Keep in mind that there are other great resources other than what is mentioned below. But these happen to be the ones I took and found useful.

PHASE 1: INFANCY

In order to be good at data science, you need to have good fundamentals in programming, statistics, and math. At a minimum, I recommend you learn the following:

University-level introduction to computer science course (For me it was C++).
University-level lower-division math courses such as multivariable calculus, differential equations, and linear algebra. This will directly impact your understanding of the low-level math of deep learning, such as backpropagation and matrix operations.
University-level introduction to statistics and probabilities that teaches you R.

The good news is that they do not have to be taken at a university. To learn the skills I mentioned above online, I recommend these resources:

Math: Multivariable calculus, differential equations, and linear algebra from Khan Academy.
Statistics: Statistics in R and Intro to data science via the Data Science Specialization by Johns Hopkins University on Coursera.
Python: CodeAcademy.com for general programming in Python.

To see examples of what data science can do, check out Kaggle.com where people learn and compete in data science projects. Also, check out DataCamp.com, which provides hands-on tutorials on various data science topics in both R and Python.

By the end of Phase 1, you should be comfortable with performing simple machine learning techniques like logistic/linear regression and decision trees on either R or Python. On a side note, I recommend learning both R and Python. Even though I mostly use Python lately, it is useful to know both depending on the problem you are trying to solve.

PHASE 2: ADOLESCENCE

Now you should have a better idea of data science and statistical methods. In Phase 2, you want to go deeper and focus on machine learning.

I found that online resources like Coursera do not usually cover as deep as a university-level course. Thankfully, Stanford’s AI Lab provides amazing courses online for free. You can watch world-class lectures, lecture notes, and many other course materials for free. I recommend you take the Coursera course and watch Stanford lectures during the same period if available. For example, DeepLearning.ai on Coursera gives you a very good and practical side of deep learning, whereas Stanford’s CS231n Computer Vision course delves much deeper.

In this phase, take the following:

Machine Learning: Andrew Ng’s Machine Learning Course on Coursera. (I took this but did not pay for the certification because the homework was not using Python or R, but it is still very useful for understanding the fundamentals).
Machine Learning: Stanford CS229 Machine Learning Course. (These are older lecture videos by Andrew Ng, but still very good).
Text Analytics: Applied Text Mining in Python on Coursera. (I have not taken this course, but text analytics and natural language processing (NLP) is a very common and desired skill for a data scientist).
PySpark: DataCamp’s Introduction to PySpark Course. (PySpark is the Python version of the Spark distributed computing framework. Simply put, it allows you to use Python on very large data. I use it on a weekly basis).
Deep Learning: Andrew Ng’s DeepLearning.ai on Coursera. (I paid for the certification because the homework on this course is very good. Since it’s very affordable, I would recommend you pay for it).
Computer Vision: Stanford CS231n Convolutional Neural Networks for Visual Recognition Course.
Natural Language Processing: Stanford CS224n Natural Language Processing with Deep Learning Course.
PyTorch and Practical Deep Learning: Fast.ai. (I have heard so many good things about this free course, and PyTorch has been gaining popularity fast).

PHASE 3: INDEPENDENCE

During this phase, you should prepare for interviews and continue to learn new and deeper topics. If you have mastered the materials up until Phase 2, you should have enough knowledge to apply for an entry-level job. However, there are a few more things that are critical for you to pass the interview process.

Personal Projects: If you are in a data science program, most classes make you complete machine learning projects, which are very good to practice your skills and to show employers what you have done. I highly suggest trying some personal projects—the easiest entry point being Kaggle. Also, even though it is not strictly necessary, I suggest having example code and completed projects on GitHub to showcase your work.
SQL: You will most likely be interviewed on SQL. When I started working at GoDaddy, I did not know too much SQL. When I was interviewing, I just learned a little bit on W3Schools.com, CodeAcademy, and by googling SQL interview questions. Even though it depends on the company, SQL is usually not as heavily weighted as your machine learning and general programming skills. It is relatively easy to learn on the job. Check out Leetcode.com to practice your SQL and programming skills.

Finally, by this stage, you should have enough knowledge to explore different machine learning topics and dive deeper. It is up to you to focus on whatever topic sounds interesting to you, whether it be RNNs, CNNs, NLP, etc.

CONCLUSION

I really hope you found this useful. I tried to focus on specific classes you should take instead of specific tools or Python/R packages you need to learn because the classes will teach you those things.

If you want to see example codes of machine learning, check out my GitHub repository, which I constantly update with new things I learn.

Thank you for reading!

Other Articles and Projects: