IBM has a whole family of certificates on Coursera and they tend to blur together, so it is worth being clear about what this one is, because it sits in a useful spot. The Machine Learning Professional Certificate is the applied, get your hands dirty option, a sequence of courses that march through the main territory of the field, supervised learning with regression and classification, unsupervised learning with clustering and dimensionality reduction, then deep learning, and even a pass through reinforcement learning, which a lot of introductory programs quietly skip. The structure I like most is that almost every concept is immediately attached to a lab in Python with scikit-learn, so you are not just watching someone describe a random forest, you are training one, breaking it, and looking at why it underperformed. By the end you have built a respectable spread of small projects, and those are the real output, more useful for a job application than the certificate line on your profile.
Where you should set expectations is depth of theory. This is firmly an applied course, so it teaches you to use the algorithms competently and to reason about when to reach for which, but it does not take you through the derivations the way a Stanford course would, and if your goal is research or you simply want to understand the maths properly you will find it thin in places. Quality is also a little uneven across the individual courses, which is common with these multi instructor IBM tracks, some modules are sharper than others. And as with most vendor certificates, a few sections lean toward IBM's own tools and platform, which is fine for learning but not always what you will be using in a job.
None of that sinks it. If you want a broad, genuinely hands on tour of machine learning that leaves you able to build things and shows up well in a portfolio, this is a solid and fairly affordable way to get there, just plan to follow it with something more rigorous if you want the theory underneath.