Before we set out to decide the path for learning data science, lets define the subject. Data Science is a discipline of applied mathematics that applies statistical methods and computer science techniques in order to understand how things work, what they are made of, and their relationships between each other. It involves solving problems. The idea behind the term “data science” is, well, it’s just data as raw material and structured around scientific study of relating elements. I think it can be said with some authority that every field has its own definition of data, but even within the same field, there are many different ways in which you might term and use data. 

Learning Data Science as a Data Wrangler

You need to learn how to use Python and R to analyze and visualize data. You'll be able to build and analyze models and visualizations using the data you collect.  will teach you the fundamentals of data science, from data analysis to data visualization, and will help you understand the difference between data wrangling and data mining. For a data scientist, it’s essential to understand the core data science fundamentals, access the data assets, organize and visualize the data, and begin to build models with the data in a systematic and organized way. This course will introduce you to the fundamentals of data analysis and data visualization. You will learn how to use data to make inferences, discover patterns, predict future events and understand causal relationships. 

This data wrangler role is perfect for someone looking to pursue a career in data science and to start doing data science with a data wrangler certification. This is a great way to get your foot in the door with the data scientist community. The Data Scientist Certification is available to anyone who wants to learn more about data and science. If you are interested in becoming an expert at your field, then this certification could be the right choice for you. It will give you the chance of working alongside experts from different fields, such as business, marketing, IT or even law.

Of course this also includes a data wrangler certification if you wish to practice data wrangling. What is the minimum skill set needed to become a data wrangler? After you successfully finish the data wrangler course certification, you are expected to learn a few key skills which are useful for both data wrangling and data science in general. These skills include: Knowledge of data structures and algorithms. This includes understanding of how data is organized and stored in databases, and how to use and manipulate data in a variety of ways. You will be required to work with your colleagues on projects that require them to understand these concepts as well as build systems around them. The best way to master any one skill would probably be through collaborative learning or an online course.

Learning Data Science as a Data Scientist

In the present era of advanced digital era, education has a serious role to play in the way data is being managed and insights derived. There are many approaches to learning data science. This can be viewed in terms of how one learns data science: from self learning through video courses; to offline learning through public databases and so on. There are different pathways that lead to data science profession and different study design requirements. Online courses offer better learning experience than their offline counterparts. Online courses are also more flexible than in-person courses. You can take them whenever and wherever you want, and you can complete them in as little time as you need. However, you need some basic knowledge of programming language before taking any online courses. Some people find it easier to learn a new language/platform if they already know the basics. In this case, you should have at least 3 years of experience in the programming languages you are interested in learning.

The Boiler Plate 

1. Start with one programming language like R or Python and then learn about a few others. Then start learning more by reading ebooks. After you have learned how to program, get out there and try your hand at it. 
2. Choose a standard statistical text book for learning the basic of statistics. You wont need inferential statistics initially unless you intend to get into research. If you are interested in learning statistics for the first time, you can start with a textbook that is a good introduction to the subject. If you want to learn more about statistics, there are many books that are available online. Statistics is one of the most popular subjects in the world. There are over 10,000 textbooks available for free on the Internet. 
3. Learn algorithms used in machine learning. My recommendation would be to cover these essential algorithms namely Random Forest, Support vector machines, Bagged trees, Boosted Trees for Supervised learning. You can leave the Computer vision and Natural Language Algorithms for now as we at a very nascent stage.
4. Look out for an unpaid or paid internship. They will add a lot of value to your profile from an experience perspective as well as credentials.
5. Connect with industry resources who can help you with mentorship as well as help you push your CV to relevant contact.
6. Pick up your first assignment on the basis of learning curve and not remuneration.

To get a detail view of the resources and certifications you can check this article.