Data Tells a Story: train delays, medical decisions, snacking smartly


We at Inform believe that data tells a story, across all industries, and every week we round up the most interesting ones. This week: predicting train delays; making wise medical decisions; and snacking smartly.

How big data predicts and helps prevent train delays in Sweden

The commuter rail operator in Stockholm is using big data to forecast and prevent delays.

Using historic data to look two hours into the future, their prediction model anticipates and acts on disruptions that have yet to happen. For example, the model may predict that a train will be 10 minutes late to a certain station. To avoid this, another train is sent to that station on time, avoiding a “ripple effect” of delays that will grow exponentially.

Big Data reveals the surprising profile of an ISIS recruit

In the light of the attacks in San Bernardino and Paris, one data scientist decided he wanted to do more than “pray and condemn the violence.”

Zeeshan ul-hassan Usmani poured over “data on ISIS recruits the way he normally analyzes data on consumers for major brands,” including social media posts and the cases of accused terrorists.

He came away with several findings. One is that there are over 70,000 people in North America, Australia, and Europe “ready to radicalize.” Another is that recruits are mostly young and male; more likely to be educated; and from middle or upper middle class families. They also don’t necessarily have a devoutly religious background but are more likely to have been secular and become radicalized.

In addition, he discovered what could be a connection between number of those ready to be radicalized and the prevalence of Islamophobia. For instance, he estimated that France has over 27,000 potential recruits (as opposed to little over 1,500 in the UK). France also has the largest Muslim prison population and has had 26 mosques vandalized since the attack at Charlie Hebdo earlier this year.

Using Big Data to Make Wiser Medical Decisions

In this article, a physician explores the different ways data can help patients better manage their health care. One way is through patient-generated data. Using data collected from a wearable device, Dr. Halamka tracked his own blood pressure levels and the possible causes, finding that his high blood pressure was most likely genetic and not caused by external factors.

Data can also help with precision medicine. When his wife diagnosed with breast cancer, Dr. Halamka was able to use open source software to assess the treatment of 10,000 women who fit his wife’s criteria and determine the best course of treatment (his wife is now cancer free).

Cruz campaign credits psychological data and analytics for its rising success

While Ted Cruz has spoken out against excessive government data collection, his presidential campaign has been actively collecting and analyzing data from supporters and potential voters to personalize messages, calls, and visits.

The data comes from a variety of sources including Facebook posts, buying habits, an app that keeps supporters “in touch” with the campaign while scraping their contacts, surveys of more than 150,000 households, and geo-fencing, geographically tracking people through their mobile devices.

From the collected data, the Cruz campaign, working with a data analytics firm, built several profiles, such as the “stoic traditionalist,” a conservative voter mainly concerned about immigration, and tailored messaging to those profiles ( “confident and warm,”  “straight to the point”).

Missives were also designed according to how people scored on certain attributes. Those who scored high on “neuroticism” would receive pro-gun messages emphasizing the use of weapons in terms of personal safety, while those who scored high for “openness” would receive a pitch on the idea of hunting as a family activity.

How Gousto is using data to change the way we shop for food online

UK-based startup Gousto makes cooking easier by delivering ingredients in a box. But they don’t just take orders: they ingest data to learn more about what their customers like.

The company built a data engine “to tag every ingredient and recipe to build up a network understanding” of their customers’ preferences. Their recommendation engine, dubbed “Laura,” analyzes millions of data points to predict what people like to eat and when.

Gousto’s tactic is similar to that of Naturebox, a U.S. startup that delivers healthy snacks and recommends snacks tailored to individual tastes based on an algorithm they developed.

Leave a Reply

Your email address will not be published. Required fields are marked *