Thank God it’s Data Stories Friday! This week: recruiting better talent; surfacing fraud; and data mining chicken tikka masala.
While employers may feel they have plenty of data points to judge a potential hire — years of experience, number of previous employers, last salary — in the end they may still be relying on a hunch about the perfect fit.
Some companies, says NPR, are offering the idea “that new behavioral science techniques” — in combination with data analytics — “can give employers more insight into hiring.”
Although intelligence and personality tests have been around for a hundred years, big data allows for the creation of “more nuanced tests” that can possibly better gauge personality traits that may “help increase productivity and reduce turnover.”
In one example, potential hires play an Angry Birds-type game while behind the scenes, the company collects thousands of data points such as level or pressure, intensity, and level of challenge, since gameplay potentially correlates with how people think and work.
LinkedIn recently released a new product that allows marketers to target ads at segments of LinkedIn’s audience and serve them ads, not just in LinkedIn but across the web.
LinkedIn follows in the footsteps of other social media companies such as Facebook, which is using login data “to help marketers connect user identities across desktop and mobile,” and Twitter, which recently launched a product “that allows its advertisers to serve targeted, Promoted Tweets to users of Flipboard and Yahoo Japan.”
LinkedIn offers the most precise targeting, “allowing marketers to pick and choose exactly the type of professionals they want to target out of its 347-million-strong network.”
In the banking hub of London, pinpointing malpractice by City traders is a priority. However, according to the Market Practitioner Panel, current methods “of monitoring for illegal trading practices, such as ‘key word surveillance’,” are flawed, and big data technology may provide a longer-term solution.
One company suggests a method called predictive coding, which “goes beyond simply looking for key words” and identifies “irregularities in patterns of behaviour,” such as communication through “unofficial channels”; working unusual hours; or missing mandatory “block leave,” or vacation.
State and city agencies have begun enlisting data companies to help root out fraud. For instance, the NYC Human Resources Administration had “data detectives” run benefit recipients through a “computerized pattern-recognition system,” which surfaced a small percentage of anomalies.
For instance, one individual had received more than $50,000 in Medicaid benefits, which raised a red flag since “most families of similar size and income typically received multiple benefits — like health coverage, food stamps and cash assistance.”
After a multisource data analysis, the data detectives found that “that the family had underreported its assets,” including an electrical contracting business, three residential properties, and banks accounts with more than $100,000.
After this kind of systematic “multisource data analysis” was implemented, staff members identified $46.5 million in fraud through almost 30,000 investigations compared with only $29 million through 48,000 investigations.
Some chefs have long believed in the food pairing hypothesis, that “ingredients that share the same flavors ought to combine well in recipes.” But while that practice is common in North American and Western European cuisines, does it apply to food from other parts of the world?
Researchers at the Indian Institute of Technology wanted to find out. By data mining more than 2,500 recipes “from eight sub-cuisines, including Bengali, Gujarati, Punjabi, and South Indian,” they found that the recipes contained 194 different ingredients, with an average of seven but as many as 40 for more “royal” dishes.
What the data told the researchers was that negative food pairings — that is, the combination of ingredients that have dissimilar flavors — dominate Indian cuisine, and that “the strength of this negative correlation is much higher than anything previously reported.” They also found that the addition of certain spices such as coriander, tamarind, ginger, and cinnamon “make the negative food pairing effect more powerful.”
All of this could potentially lead to the creation of even more novel Indian dishes.