NYU Data Science newsletter – September 28, 2015

GROUP CURATION: N/A

… While some of you might be busy going through the 12 Best Free Ebooks for Machine Learning, chances are that many of you have taken the step and trying to gather the resources and learning ways to proceed and handle projects. If you are the one who has decided to pursue then we promise to help you with as much useful stuff as possible. Today, we have come up with best machine learning cheat sheets that are not too much in number but are simply worth it.

Take a look and enjoy the ease of workflow with these below listed best machine learning cheat sheets.

ORCID provides a persistent digital identifier that distinguishes you from every other researcher and, through integration in key research workflows such as manuscript and grant submission, supports automated linkages between you and your professional activities ensuring that your work is recognized.

APIs — the rules governing how software programs interact with each other — not user interfaces, will upend software for years to come.

When Intel CEO Brian Krzanich doubled down on the Internet of Things at the company’s annual Developer Forum in August, he emphasized what many of us have already known — the dawn of a new era in software engineering. It’s called API-first design, and it presents a tremendous opportunity for developers who adapt — not to mention a major risk for developers (and companies) who don’t.

The American Red Cross is stitching thousands of crowdsourced photos into OpenStreetMap to get a better view of the world’s most vulnerable communities and assess their needs as part of its Missing Maps project.

With their market onslaught in the past few years, fitness wearables are easily dismissible as a fad. But some data from research firm Gartner deflates that notion, with a prediction that the number of wristbands, smartwatches and other such devices shipped to consumers will triple in the next five years to 200 million units, reaping nearly US$16 billion in sales.

As much as consumers are expanding the health and wellness market, professional sports teams are reaching into their wallets to beef up their athletes, not just with wearables but with a broad range of sports technology.

And as data analytics become more mainstream in sport, they are playing a role to shape the game, along with the burgeoning class of wearables that fine-tune workouts. At the forefront of this movement is a new breed of startups — including Sportlogiq, Stathlete and PUSH — hoping to cash in on this trend.

… Research data management (RDM) has been hotly debated in recent years, yet it is a concept with a remarkable simple premise: the organization and storage of data. However, a critically important facet of this is future-proofing data collation systems so that material remains useable for as long as possible, and as such RDM requires planning both for current needs, and those of an unknown future.1-3 Most funders require researchers to submit a data management plan with their grant applications, and many Research Councils have their own policies and principles that must be adhered to. On top of this, there are legal requirements to consider, such as freedom of information legislation and individual rights to privacy and confidentiality. Data management has therefore become a complicated issue, and with data outputs growing at an exponential rate, it is one that needs to be addressed by anyone involved in generating, using and sharing research information.

Current thinking is pointing to one key truth at the heart of RDM: libraries and librarians play a central role, and can bring enormous value and insight to the process – particularly around the use of data repositories, as well as in defining standards for data description, accuracy and accessibility.

Watching another episode on Netflix, reading the Guardian online and downloading apps are not obvious ways to pollute the atmosphere. But collectively, our growing appetite for digital services means the datacentres that power them are now responsible for about 2% of global greenhouse gas emissions, a similar share to aviation.

In his keynote speech at Med X, Dr. Eric Topol of the Scripps Institute delivered a plea for medicine democratization through data transparency.

ixty years ago, statistics barely touched the school experience of a typical student. In the study of social science, students might encounter data. In a science laboratory experience, students might collect data. In a mathematics classroom, students would be expected to know how to calculate the mean of a set of numbers. In contrast, today it is becoming prevalent to expect increasing numbers of students to learn several measures of central tendency and spread, to encounter theoretical and actual distributions, and to discuss topics such as randomness, statistical tests, and statistical significance that in the past were introduced at the college level. As one of the mathematical sciences, the study of statistics in grades K–12 naturally has been considered as a part of the school mathematics curriculum. This has great advantages, as mathematics is the second most important academic school subject, behind reading and language arts. But as statistics has become more important, its connections with everyday literacy, science, health, and the social sciences suggest teaching statistics across the curriculum in addition to a reconsideration of its relationships with mathematics.

… he and his team are working to use data to improve civilian life in a number of ways. In health care, they’re working on harnessing genomic and personal health information to work on precision medicine, which the initiative describes as “an innovative approach to disease prevention and treatment that takes into account individual differences in people’s genes, environments, and lifestyles.”

In criminal and social justice, Patil wants to improve policing by encouraging different states and precincts to share information in order to compare stop rates, search rates, and other metrics to “[give] data back to the officer to make better decisions.” Patil says projects are also underway in in higher education and housing development.

In this Advanced Machine Learning with scikit-learn training course, expert author Andreas Mueller will teach you how to choose and evaluate machine learning models. This course is designed for users that already have experience with Python.

You will start by learning about model complexity, overfitting and underfitting. From there, Andreas will teach you about pipelines, advanced metrics and imbalanced classes, and model selection for unsupervised learning. This video tutorial also covers dealing with categorical variables, dictionaries, and incomplete data, and how to handle text data. Finally, you will learn about out of core learning, including the sci-learn interface for out of core learning and kernel approximations for large-scale non-linear classification.

Once you have completed this computer based training course, you will have learned everything you need to know to be able to choose and evaluate machine learning models. Working files are included, allowing you to follow along with the author throughout the lessons.

… That future robots will have human-like emotions is a huge misconception, said Yann LeCun, the director of Facebook’s Artificial Intelligence Research team.