General Applications Product Guide: Fill & Download for Free

It depends on your objective. C++ has problems, especially for a beginner, and it is often a poor choice for general applications, business and web programming.The world of languages and paradigms is much wider and more interesting than C#, C++, Python, and Objective-C (all languages of the 1980s and 1990s, not so much the future). You will probably learn at least 6-10 languages in a career. Don't look at the languages that were popular in the last 30 years; look for what is relevant to in the next 30 years. C++ will continue to be around, but it isn't modern, productive or expressive enough to dominate.A bit of history: C was invented nearly 38 years ago as a way of writing the Unix kernel in a less tedious way than using PDP-11 assembly language. Programming language R&D hasn't stood still in those 38 years—and it has been guided by the need for programmers to write programs more correctly and more productively.

Unfortunately, data science interviews are highly variable company-to-company. While I'm not a Google employee, I assume Google is no exception. From my understanding from friends, there are two paths into Google as a data scientist.The first is a specialist application or referral. These applicants typically have already built popular software systems or published papers in the field relevant for this team. For example, the Google Brain team focuses hires from individuals with publications in ICML or NIPS. If this is you, you should already be prepared for the interview.The second is a general applicant; here you will be asked canonical software engineering and statistics questions. Here, I recommend you take a look at the following:Software Engineering Cracking the Coding Interview is the single greatest source for typical software engineering questions. It may sound contrived, but practice coding solutions both on your computer and on a whiteboard. It's worth the $15 for a whiteboard in your own home.Statistics I recommend you take a look at two resources. The first, The Unofficial Google Data Science Blog . This blog contains a nice overview of A/B testing and random sampling from current Google employees. The second, A Practical Guide To Quantitative Finance Interviews. While this guide is focused on Wall Street quantitative finance, the probabilistic problems are frequently used across data science.Machine Learning The goal is to know common algorithms in machine learning (SVM, random forest, neural networks, naive bayes, ...) and common pitfalls in machine learning (train/test split, overtraining, gradient shrinking / explosion). It's difficult to recommend any one resource for learning machine learning. Hopefully if you're interviewing for a data science position you're already familiar with a lot of these principles. If not, check out Kaggle.Product This last category is the most amorphous. Some companies care more than others whether data scientists are intrinsically product focused individuals (e.g., given a market, how would you build a product?). Here I recommend Case Interview Secrets. Again, the focus is on a different job (consulting) but the frameworks around which Victor analyses businesses is very useful for identifying opportunities in markets.

It does depend on what you are trying to achieve.Indeed, for nearly all general programming, C is no longer appropriate, simply because we have better, more productive languages that make it much easier to write correct programs. In fact, what has been learned in the 40 years since C was invented, is that there is nothing special about the C front end when it comes to performance (this is just common sense, really). You can generate efficient machine code from more expressive front-ends that, rather than taking a perverse pride in lacking abstraction, provide efficient aids to correctness (such as proper types, and closures).But in the 1980s and 1990s C was considered a general applications programming language—resulting in a large mass of legacy C codebases that distort popularity and relevance statistics :)It's not entirely obsolete, but its niche of applicability is steadily shrinking. It is going to remain the go-to language (get it?:) for embedded and small devices for a while yet, for some practical reasons, as Diemisom Melo's answer says."Popularity" is a worthless guide to questions like this.