The Ultimate Guide on the Data Science MicroMasters Programs on edX 2021 (2023)

Which of the 6 programs should you choose?

Updated 2021–03–14: The MicroMaster Artificial Intelligence from Columbia University is no longer available, and no further runs are planned.

Massive Open Online Courses (MOOCs) are today a valuable alternative for learning data science.

Not only beginners but also data science veterans are doing MOOCs — like myself.

I am a part-time lecturer on master’s level programs and familiar with many lectures worldwide, and I regularly assess online courses.

Not a few students do in parallel MOOCs in the data science field. These are mostly students from other disciplines that are moving into that area, and for them, it is a fast and flexible way to fill the gap in basics while attending lectures on the master’s level.

So, they often ask me for advice which MOOCs would be best suited for their individual situation.

There are pros and cons to each MOOC. The fact is that with MOOCs alone, you cannot become a data scientist. You need practical experience and exchange with and mentorship from veterans. But it is a convenient way to get access to high-quality education and a fast path to fill gaps.

My observation is that the MOOCs’ quality is increasing steadily, and you have a growing choice of online programs. So, it becomes essential to find the individually right program.

Another question is if you should get a certificate or not. My personal opinion is that when you anyway will get a degree on a graduate level, and it only serves as a possibility to fill gaps, it is not necessary to take a certificate. If it shows additional knowledge that you would otherwise not have, go for a certificate. Benjamin Obi Tayo, Ph.D. has an excellent summary of the consideration for certifications.

Comparing the many different platforms is not a simple task; even the comparison of the many various programs on one platform is tricky for a data science beginner.

Our university is part of the edX platform, so it is the choice of many of our students. For that reason, I am most familiar with the course offered there.

In this guide, I focus on the so-called MicroMasters programs on the edX platform.

A MicroMasters program consists of a series of graduate-level courses, including corresponding assignments, exams, and practical projects. The average length is about one year, with an effort of about 10 hours per week. It corresponds to around 25% of a full master’s program, and the earned credits can be transferred to designated online or on-campus master’s programs.

So, MicroMasters programs are not only a convenient way of gaining relevant and high-quality education but also a way of testing your delight in full graduate education.

The following six MicroMasters on edX programs have been assessed:

1. Analytics: Essential Tools and Methods — The Georgia Institute of Technology
2. Artificial Intelligence — Columbia University — no longer available (see below) [updated 2021–03–14]
3. Big Data — The University of Adelaide
4. Data Science — UC San Diego
5. Predictive Analytics using Python — The University of Edinburgh
6. Statistics and Data Science — MIT Massachusetts Institute of Technology

The programs are listed in alphabetical order.

Remarks:

• All information below is taken from the corresponding MicroMasters program description on edX, the individual course descriptions, and from the courses themselves only accessible when enrolled.
• The “Pros,” “Cons,” and “Who should take that program” represent my opinion based on my assessment of the programs and courses.

1. Analytics: Essential Tools and Methods — The Georgia Institute of Technology

What you will learn: You learn data preparation, regression, classification, clustering, time series models, prediction, optimization and model validation, preprocessing unstructured data, relational data, tidying data, visualization, numerical computing, customer analytics, measuring risk and returns, marketing analytics, statistical process control, and forecasting demand.

Level: Advanced

Type: Instructor-led

Courses: 3 courses:

• Introduction to Analytics Modeling (16 weeks, 8–10 hours per week)
• Computing for Data Analysis (15 weeks, 8–10 hours per week)
• Data Analytics for Business (16 weeks, 10–12 hours a week)

Prerequisites:

• Probability and statistics
• Basic programming proficiency
• Linear algebra
• Basic calculus

Lengths and effort: 1 year, 9–11 hours per week

Assessments and certification: A grade of 60% or higher is required to pass, from weekly homework/notebook, a project, and proctored exams.

Tools and programming language(s): SQL, R/RStudio, Python, Arena simulation software, PuLP optimization software, Simply, pandas, Numpy, Scipy

Costs: USD 1,500 (full program)

Pros: Gives a good introduction in the most important methods and tools, including the link to business data analytics problems.

Cons: Focuses on classical optimization with specialized tools rather than advances in machine learning and deep learning.

Credits: The MicroMaster counts with 25%, or 9 hours toward the total of 36 hours of the Online Master of Science in Analytics (OMS Analytics) degree of the Georgia Institute of Technology.

Who should take that program: People who want a sound introduction into data analytics with the goal to apply it in business and finance.

2. Artificial Intelligence — Columbia University

Remark: the program is no longer available and no further program runs are planned [updated 2021–03–14]

What you will learn: You learn AI applications of machine learning, reinforcement learning, natural language processing, regression, classification, boosting, unsupervised learning, Hidden Markov Models and State-space Models, principle component analysis, robotic movements and kinematics, robots motion planning, all aspects of computer-generated imagery programming and creation.

Level: Advanced (graduate-level course)

Type: Instructor-led

Courses: 4 courses:

• Artificial Intelligence (AI) (12 weeks, 8–10 hours per week)
• Machine Learning (12 weeks, 8–10 hours per week)
• Robotics (10 weeks, 8–10 hours per week)
• Animation and CGI Motion (12 weeks, 8–10 hours per week)

Prerequisites:

• Proficiency python programming skills
• Proficiency programming knowledge in C/C++
• Calculus incl. partial derivatives and linear algebra
• Introductory classical mechanics
• Probability and statistical concepts

Lengths and effort: 1 year, 8–10 hours per week

Assessments and certification: Grade of 60% or higher required to pass, of quizzes, (programming) projects, homework, and the final proctored exams per course.

Tools and programming language(s): Python, C/C++

Costs: USD 896.40 (full program)

Pros: One of the very few graduate-level MOOCs that also provides a more in-depth introduction to robots and computer-generated imagery.

Cons: Even though the course descriptions note a weekly effort of 8–10 hours, that only applies to people fulfilling the prerequisites on a proficient level. All others should consider 10–20 hours a week to complete the courses successfully.

Credits: It will count 25% or 7.5 credits toward the 30 credits of the Master of Computer Science program at Columbia University.

Who should take that program: People proficient in programming in Python and C/C++, and with an advanced knowledge in data science who want to deepen the AI skills on a graduate level.

3. Big Data — The University of Adelaide

What you will learn: You learn algorithm design, and fundamental programming concepts, manipulation, and transformation of data, as well as dimension reduction, directed and undirected graphs, hashing, big data fundamentals, stream processing, MapReduce, Google’s web search, and AdWords system, regression and classification, and deep learning algorithms.

Level: Introductory to advanced

Type: Self-paced

Courses: 5 courses:

• Programming for Data Science (10 weeks, 8–10 hours per week)
• Computational Thinking and Big Data (10 weeks, 8–10 hours per week)
• Big Data Fundamentals (10 weeks, 8–10 hours per week)
• Big Data Analytics (10 weeks, 8–10 hours per week)
• Big Data Capstone Project (6 weeks, 4–5 hours per week)

Prerequisites: None

Lengths and effort: 1 year, 8–10 hours per week

Assessments and certification: Grade of 60% or higher required to pass, based on all quizzes, assignments, and worked problems.

Tools and programming language(s): Processing.js, R/RStudio, Java/Eclipse IDE, Apache Spark, sparklyr, and H20.

Costs: USD 1,436.40 (full program)

Pros: It gives a broad and comprehensive introduction into the Big Data thinking and design, the various tools and programming languages, and integrates real-world examples from web search and web advertising.

Cons: It is less suited for a data science beginner; otherwise, the effort will be considerably higher.

Credits: The Big Data MicroMasters program certificate represents 25% or 12 units of the Master of Data Science program, which requires a total of 48 units of coursework to complete.

Who should take that program: People who already have a sound understanding of data science and are familiar with at least one programming language on an intermediate level and want to have an introduction into the Big Data area.

4. Data Science — UC San Diego

What you will learn: You will learn the necessary process of data science, Python and Jupyter notebooks, processing uncurated datasets, fundamental statistical analysis, basic machine learning methods, visualization, an overview of sets, counting principles, combinatorics, discrete probability, conditional probability, and Bayes’ Rule, random variables, expectation, variance, and correlation, joint distribution families, continuous distributions, probabilistic inequalities, concentration, and limit theorems, linear and logistic regression, sampling, parameter estimation, and confidence intervals, hypothesis testing, decision trees, and an introduction to deep learning, neural networks, and TensorFlow.

Level: Advanced

Type: Self-paced

Courses: 4 courses:

• Python for Data Science (10 weeks, 8–10 hours per week)
• Probability and Statistics in Data Science using Python (10 weeks, 10–12 hours per week)
• Machine Learning Fundamentals (10 weeks, 8–10 hours per week)
• Big Data Analytics Using Spark (10 weeks, 9–12 hours per week)

Prerequisites:

• Programming experience on an undergraduate level
• Logic and set theory
• Multivariate calculus
• Linear algebra

Lengths and effort: 10 months, 9–11 hours per week

Assessments and certification: Grade of 70% or higher required to pass, from quizzes, projects, final exams in the first course, and 65% or higher in the second, third and fourth course.

Tools and programming language(s): Python, Jupyter notebooks, pandas, NumPy, Matplotlib, git, scikit-learn, NLTK, Spark, and PySpark

Costs: USD 1,260 (full program)

Pros: A fundamental introduction for all data science beginners covering many different topics, including a sound introduction into statistical concepts. Letter grading is offered.

Cons: The program gives a broad but not necessarily in-depth knowledge. Technical deepness must be studied by yourself and can take a lot of effort.

Credits: The MicroMaster will count as 30% of the full Master of Science degree in Data Science of the Rochester Institute of Technology. It also counts 25% (or 100 credits of the 400 credits) towards the Master of Predictive Analytics at Curtin University. Finally, completers will receive a favorable review of their application if you choose to apply to the Online Master’s Degree program in Data Science from UT Austin on edX.

Who should take that program: This course is designed for beginners in data science to get an overview from statistical concepts to the most used models and gives a flavor on deep learning and big data.

5. Predictive Analytics using Python — The University of Edinburgh

What you will learn: You learn basics of the predictive modeling life cycle, introduction to Python, preparation of data, data quality, under- and oversampling, training, validation, confidence intervals of estimators, linear and logistic regression, decision trees, random forest, neural networks, overfitting, and pruning.

Level: Advanced (graduate level)

Type: Instructor-led

Courses: 5 courses:

• Introduction to Predictive Analytics using Python (6 weeks, 8–10 hours per week)
• Successfully Evaluating Predictive Modelling (6 weeks, 8–10 hours per week)
• Statistical Predictive Modelling and Applications (6 weeks, 8–10 hours per week)
• Predictive Analytics using Machine Learning (6 weeks, 8–10 hours per week)
• Predictive Analytics Final Project (6 weeks, 8–10 hours per week)

Prerequisites:

• Undergraduate level of mathematics and statistics
• Some previous experience in programming like Python, Java, or C/C++

Lengths and effort: 8 months, 8–10 hours per week

Assessments and certification: A grade of 50% or above is required based on multiple choice and coding assessments; it requires a final project consisting of a theory-based written exam, a Jupyter notebook submission that reflects a real-life case study, and a 1,700-word reflective submission based upon the Jupyter notebook submission.

Tools and programming language(s): Python, NumPy, pandas, scikit-learn

Costs: USD 1,350 (full program)

Pros: A very sound technical introduction into modeling, model validation, sampling, and model comparison with a strong emphasis on data examples and programming.

Cons: The requirement of a grade of 50% seems low, but the effort especially, for absolute beginners in that field, will be higher as there is a strong emphasis on practical applications and programming.

Credits: When pursuing a corresponding master at the University of Edinburgh, 30 credits are awarded towards the 180 credits required for a full master.

Who should take that program: People who want to have technical depth in predictive analytics and develop strong programming skills than just a broad overview of many topics.

6. Statistics and Data Science — MIT Massachusetts Institute of Technology

What you will learn: You learn about probability models, Bayes’ rule, independence, discrete and continuous distributions, Bayesian inference, classical statistics, random processes, Markov chains, estimations, regressions, A/B testing, experimental design, data visualization, method of moments, and maximum likelihood, confidence interval, hypothesis testing, goodness of fit tests, predictive models, principal component analysis, overfitting, regularization and generalization, clustering, classifications, recommender problems, probabilistic models, EM algorithm, reinforcement learning, support vector machines, neural networks, and deep learning.

Level: Advanced

Type: Instructor-led

Courses: 5 courses:

• Probability — The Science of Uncertainty and Data (16 weeks, 10–14 hours per week)
• Data Analysis in Social Science — Assessing Your Knowledge (4 weeks, 10–14 hours per week) — prerequisite is first a passing grade in the course Data Analysis for Social Scientists by the MIT on edX (11 weeks, 12–14 hours per week)
• Fundamentals of Statistics (18 weeks, 10–14 hours per week)
• Machine Learning with Python: from Linear Models to Deep Learning (15 weeks, 10–14 hours per week)
• Capstone Exam in Statistics and Data Science (2 weeks, 10–14 hours per week)

Prerequisites:

• Single-variable and multivariable calculus
• Mathematical reasoning
• Familiarity with sequences, limits, infinite series, the chain rule, and ordinary or multiple integrals
• Vectors and matrices
• Proficiency in Python programming on a level covered by the course Introduction to Computer Science and Programming Using Python by the MIT on edX

Lengths and effort: 1 year and two months, 10–14 hours per week

Assessments and certification: Grade of 60% or higher is required based on lecture exercises, problem sets, homework, projects, and exams.

Tools and programming language(s): R, Python

Costs: USD 1,350 (full program)

Pros: It is a very comprehensive and rigorous program that teaches the mathematical and theoretical foundations of statistics and data science. Also, more than 20 universities around the globe count the credits towards their master’s programs.

Cons: The program is very academically driven, and there is less focus on the tools used in practice and the bridge to practical data science work. The prerequisites require additional courses not covered by this MicroMasters program.

Credits: Many universities around the world are transferring the credits of this program to their graduate programs. You can find the long list and all details on MIT’s dedicated webpage.

Who should take that program: People who aim for a rigorous mathematical and theoretical basis, and want to pursue a full master’s degree, are right with this program. Further, the credential of the name MIT opens the door for many other programs.

There are programs for any level of experience: for the data science beginner who either wants to have a broad or a profound technical introduction and the experienced veteran who wants to gain specialized knowledge.

The quality and the needed effort are comparable in all programs.

The purpose should guide your choice.

• Do you want to go for a full master’s program?
• Or, do you want to make your first steps in the data science field?

The program alone does not make you a data scientist. It provides you with a fundament. Then, it is upon you what you build on it.

Do you like my story? Here you can find more.