Alternatives to Coursera for Machine Learning Classes, On and Off the Web

Written by Dr Debashis Guha, Director of Machine Learning Program at SP Jain School of Global Management.

As the disciplines of Machine Learning and Artificial Intelligence become more and more important in the workplace, and demand outstrips supply for these skills, a lot of technology workers are looking to acquire these skills by taking courses, often from web based education portals and MOOCs, i.e., Massive Open Online Courses.

The leader among these web-based portals for Machine Learning classes is Coursera, which offers online courses in several disciplines from some of the world’s best-known universities. Coursera was started by Professor Andrew Ng based on his experience of teaching an online version of his Stanford Machine Learning class. Prof Ng’s Machine Learning course has been offered through Coursera since 2012 and remains the leading such online offering.

A highly rated alternative to Coursera comes from another major US university, Columbia. Taught by Professor John W. Paisley, this course is offered through EdX, a Coursera rival. This course is more comprehensive than the Coursera offering, and it has received very high ratings from students. Another EdX course that merits attention comes from Microsoft and this one is more focused on implementation issues.

The Machine Learning course from Georgia Tech offered through Udacity, yet another MOOC portal, is also highly rated by students. This class is based on one that is offered at Georgia Tech where it is a part of the Online Master’s Degree (OMS), but taking it online does not earn credit towards the OMS degree.

The online education portal Udemy has several good courses on Machine Learning. One of the best is the course titled Machine Learning A-Z™: Hands-On Python & R In Data Science, created by a team of industry experts. This course offers instruction in both Python and R as well as a comprehensive introduction to Machine Learning. Another highly rated course available on Udemy comes from the group that produces the Lazy Programmer series of books and courses. Fees for Udemy courses fluctuate widely, and are about ₹640 on average.

An interesting brief course with a focus on big data is offered by Queensland University of Technology through the portal FutureLearn. This course lasts only three weeks.

Most of the offerings above are single courses, with a Certificate of completion added on for a cost. In addition, Coursera also offers a Master’s degree in Data Science from the University of Illinois that consists of 25 courses and requires taking 15-20 hours of classes for 18-36 months. Udacity offers what it calls a nanodegree in Machine Learning. This consists of three courses, covering foundations, basics and advanced topics, and is spread out over 10-12 months.

Although such online instruction has become quite popular of late, traditional classroom teaching still has a big role to play for learning demanding subjects like Machine Learning. Online classes have advantages such as flexibility, ease of repetition, scaling, and higher availability of the best teachers. However, inadequate attention and involvement in online settings have often led to extremely high rates of dropout. Classroom instruction produces better attention and involvement, and fosters better interaction among faculty and peers.

The method of “blended learning” tries to merge the advantages of both types of learning by combining the use of internet and digital media with traditional classroom teaching. Students go over instructional materials such as texts, video and streaming apps, on their own. The material to be studied at home typically includes theoretical concepts and background matter that the students can learn at their own pace and schedule. This is followed by face-to-face sessions, often in a laboratory or workshop type setting, that are focused on building applications and exercises for practical use. These classroom sessions typically include project work set in a Lab, that is supervised and mentored by faculty, and this is critical to a thorough learning of the subject.

The advantages of blended learning have led to its being adopted by many leading educational institutions. The Chronicle of Higher Education reported in 2015 that both Yale and Johns Hopkins have adopted the blended method for many classes. Forbes magazine reported in 2016 that Imperial College London has started a pilot program in blended learning and speculated that this could be the future of all higher education. Blended learning has also been adopted at SP Jain Global. With the part-time program in Machine Learning, SP Jain Global aims at making the course accessible to working professionals while also enabling classroom discussions and personal interactions with the faculty.

Empirical analysis suggests that blended methods produce better learning outcomes than either purely digital or purely classroom methods. For instance, in a meta-analysis of empirical studies published in the prestigious journal, Teachers College Record, a team from SRI International found that
… on average, students in online learning conditions performed modestly better than those receiving face-to-face instruction. The advantage over face-to-face classes was significant in those studies contrasting blended learning with traditional face-to-face instruction but not in those studies contrasting purely online with face-to-face conditions

In another study, published in Frontiers in Psychology, two academics from Troy University have studied the transformative potential of blended learning on higher education. They write:
Students can now design flexible, life-balanced course schedules. Higher knowledge transfer rates may exist with blended course formats with online quizzes and valuable class time set for Socratic, quality discussions and creative team presentations.

Blended learning combines the advantages of digital methods such as flexibility and repetition with the high levels of attention, involvement, and interaction associated with classrooms and thus can have the best of both worlds.

About the Author: Dr. Debashis Guha 

Dr. Debashis Guha has more than two decades of experience in data analytics and machine learning, with an emphasis on economic and financial applications. He has taught at Texas Christian University, University of Texas – Dallas, Columbia University – USA, and at Administrative Staff College – India. Dr. Guha has published many articles in top rated peer reviewed journals and has been a featured speaker at international conferences. He has also been quoted by the Wall Street Journal, the Los Angeles Times, and has appeared on CNN. Dr. Guha is the founder and CEO of a Bangalore based data analytics company that specialises in
data science and machine learning for businesses and financial companies.
He has a PhD from Columbia University in New York, a Master’s from Texas Christian University, and a Bachelor’s from Indian Institute of Technology, Kharagpur.