There are many great publishers and books on Data Science. I would like to mention the ones I have benefitted from the most.
The first one is Manning Publishing. What I like about their books is how well the chapters are organized. At least the ones I have read, all had similar structures, good code listings with comments, tables and screenshots.
Here is a great introductory level book using Python - Introducing Data Science, 2016 It covers the main concepts of data science, machine learning, distributed computing, NoSQL and Graph databases, text analytics, data visualization and so on.
Just read this preface.
It’s in all of us. Data science is what makes us humans what we are today. No, not the computer-driven data science this book will introduce you to, but the ability of our brains to see connections, draw conclusions from facts, and learn from our past experiences. More so than any other species on the planet, we depend on our brains for survival; we went all-in on these features to earn our place in nature. That strategy has worked out for us so far, and we’re unlikely to change it in the near future.
But our brains can only take us so far when it comes to raw computing. Our biology can’t keep up with the amounts of data we can capture now and with the extent of our curiosity. So we turn to machines to do part of the work for us: to recognize patterns, create connections, and supply us with answers to our numerous questions. The quest for knowledge is in our genes. Relying on computers to do part of the job for us is not—but it is our destiny.
Doesn't it sound like Dataism manifesto?
The other publisher I like is Packt. They have a great variety content and their books do not all look alike. They have great deal on eBooks.
If I had to pick just one book to explain Machine Learning, I would undoubtedly pick Python Machine Learning, 2017 by the great Sebastian Raschka and Vahid Mirjalili. After getting through a whopping 500+ pages I had only one complaint - I wish it were longer!
Great content covering from the simplest algorithms to complex neural networks, supervised and unsupervised learning. Explains well what embeddings are and how back propagation works - hint: there is no sorcery involved. Uses scikit-learn and Tensorflow, and even walks you through how to create a simple predictive web app using Flask. I wish they follow up with a similarly great book on Reinforcement Learning.