NYU Data Science newsletter – May 18, 2015

GROUP CURATION: N/A

Researchers use propensity score matching (PSM) as a data preprocessing step to selectively prune observations prior to applying a model to estimate a causal effect. The goal of PSM is to reduce imbalance in pre-treatment covariates between the treatment and control groups, thereby reducing the degree of model dependence and potential for bias. Although some applied researchers have combined PSM with various ad hoc procedures and checks to produce useful analyses, we show that the core PSM procedure itself often accomplishes the opposite of what is intended — increasing imbalance, model dependence, and bias. The weakness of PSM is that it approximates a completely randomized experiment, rather than, as with other matching methods, a more powerful fully blocked randomized experiment. PSM is therefore blind to the portion of imbalance that would have been eliminated by also approximating full blocking.

Welcome to Toolbox: Nature’s hub for scientific software, apps and online tools.

Similar to data munging described in my previous post, enterprises must clean their data to derive the best possible insights for the business. Data cleaning is part of the “transform” step in the extract, transform, load (ETL) process. Arguably the most tedious of all the steps in the data manipulation pipeline, cleaning is the act of fixing spelling errors, deleting erroneous white spaces, removing duplicate entries and performing any other action to eliminate inaccurate entries in a data set. With big data containing millions of entries, addressing these issues can be prohibitively time consuming.

What happens if a company decides not to clean its data? Embarrassing mistakes are a best-case scenario; worst case, executives draw incorrect conclusions about the business and act on them.

Thankfully, there are a number of data cleaning programs that when used correctly can reduce a month-long project to mere days. This is a fast growing market, and here are a few new tools that have entered recently.

… today’s data platforms range greatly in complexity. From simple caching layers or polyglotic persistence right through to wholly integrated data pipelines. There are many paths. They go to many different places. In some of these places at least, nice things are found.

So the aim for this talk is to explain how and why some of these popular approaches work. We’ll do this by first considering the building blocks from which they are composed. These are the intuitions we’ll need to pull together the bigger stuff later on.

There’s been lots of buzz about “statebin” maps of late. A recent tweet by @andrewxhill referencing work by @dannydb pointed to a nice shapefile that ends up being a really great way to handle statebin maps (and I feel like a fool for not considering it for a more generic solution earlier).

… the Allen Institute for Brain Science, a key player in the BRAIN Initiative, has launched a database of neuronal cell types that serves as a first step toward a complete understanding of the brain. It’s the first milestone in the Institute’s 10-year MindScope plan, which aims to nail down how the visual system of a mouse works, starting by developing a functional taxonomy of all the different types of neurons in the brain.

Commonwealth Computer Research Inc. specializes in data science and analysis. The Charlottesville company hit the big leagues of technology landing a partnership with the multibillion-dollar company Google. … Years in the making, Google just released Cloud Big Table to the public. It allows storage for large amounts of spatial data, which deals with geographic location.

Sea scallops, one of the most valuable commercial fisheries in the United States, are a well managed and monitored fishery, yet little is known about how changing ocean temperatures and ocean chemistry and other environmental factors could impact the fishery. A new study published May 6 in PLOS ONE describes a new computer model to help inform scallop management discussions and decisions in the coming decades.

Researchers from the Woods Hole Oceanographic Institution (WHOI), NOAA Fisheries’ Northeast Fisheries Science Center (NEFSC), and Ocean Conservancy developed an integrated assessment model (IAM) to reproduce scallop population dynamics, market dynamics and seawater chemistry between 2000 and 2012. Data on actual landings, revenue, biomass, number and scallop size distribution were provided by NOAA’s Northeast Fisheries Science Center (NEFSC), along with ocean temperature, salinity and other oceanographic information.

A growing number of artists are using data from self-tracking apps in their pieces, showing that creative work is as much a product of its technology as of its time.

Here’s a list of universities with arguably the greatest computer science programs: Carnegie Mellon, MIT, UC Berkeley, and Stanford. These are the same places, respectively, where Andrew Ng received his bachelor’s degree, his master’s, his Ph.D., and has taught for 12 years.

Ng is an icon of the artificial intelligence world with the pedigree to match, and he is not yet 40 years old. In 2011, he founded Google Brain, a deep-learning research project supercharged by Google’s vast stores of computing power and data. Delightfully, one of its most important achievements came when computers analyzing scores of YouTube screenshots were able to recognize a cat. (The New York Times’ headline: “How Many Computers to Identify a Cat? 16,000.”) As Ng explained, “The remarkable thing was that [the system] had discovered the concept of a cat itself. No one had ever told it what a cat is. That was a milestone in machine learning.”

Parents will do just about anything to keep their children safe. Ironically, parents may also be the primary reason their children are vulnerable online. A new study suggests that parents’ social-media postings can put young children at risk, due to what seem like fairly innocuous activities. It turns out that compromising children’s safety online doesn’t take a hacker — just a savvy individual and some information that parents are all too willing to provide.

This news comes by way of a research paper entitled “Children Seen but Not Heard: When Parents Compromise Children’s Online Privacy” issued by New York University Polytechnic School of Engineering in Brooklyn. The paper asserts that parents who share their children’s information on social media are unwittingly putting their children and their families at risk by compromising personal privacy and, by extension, security.