Adaptive Behavioral Analytics: Software That Predicts the Future From Normal Patterns

The whole point of big data is, at its essence, to provide users with the ability to take normal behavior patterns and extrapolate in order to predict future behavior that might help someone make money - or help prevent them from losing it. 

That a company called Featurespace is explicitly touting its ability to do this speaks to the increasingly crowded nature of the data analytics market. The following article provides evidence of the potential pitfalls.

One of the investors in this new enterprise is the founder of Autonomy, a company acquired by HP for a handsome price, only to find itself on the business end of a substantial lawsuit when the unhappy acquirer belatedly wrote down $8.8 billion of the $11 billion after discovering what it alleges were accounting misstatements.

Given the equal attention paid to the fraud charges as well as the new company's adaptive behavioral prospects, one can only conclude that the software, to say nothing of the journalism, infers that the investor's previous actions could well be normal for him and therefore predictive of future unpleasant surprises. How useful - and inconvenient. JL

John Aglionby reports in the Financial Times:

A new form of data analysis called “adaptive behavioural analytics”, can predict future behaviour based on normal patterns of behaviour.
Mike Lynch, the former chief executive and founder of software company Autonomy, is among a group of investors in Featurespace, a Cambridge company that uses algorithms to analyse and predict behaviour.

Mr Lynch is in a long-running dispute with Hewlett-Packard over who is to blame for the multibillion-dollar losses the US company suffered when it bought Autonomy in 2011. He is a member of the Cambridge Angels group who, along with Nesta, the innovation charity, are investing £1.3m in Featurespace.

Among Featurespace’s products is its Fraud Manager, which helps companies “spot new types of fraud as it occurs, in real time, thereby allowing organisations to reduce risk and operations costs while improving the customer experience”, according to a company statement.
HP’s $11bn purchase of Autonomy in 2011 was marred a year later when the US company wrote down the transaction by $8.8bn. Some $5bn of the writedown was due to accounting misstatements, HP said.
Mr Lynch, a Cambridge university scientist who founded and ran Autonomy, has claimed that the disputed transactions reflect a difference in accounting treatment between the US and the UK.

He has accused HP of misleading its shareholders, claiming it “selectively leaked documents and information to the international media, frequently using material taken out of context to create false impressions and smear our reputations”.

Mr Lynch first invested in Featurespace in 2008, according to Martina King, chief executive, which is when it was converted from a consultancy into a technology company. It now employs 20 people.

Another of the Cambridge Angels investing in Featurespace is Andrew Hodson, a former investor in DeepMind Technologies, the UK artificial intelligence start-up sold to Google in January for £400m.

Featurespace says it is pioneering a new form of data analysis called “adaptive behavioural analytics”, which can predict future behaviour based on normal patterns of behaviour.
“Simply put, we predict the future,” the company boasts on its website, although Ms King could not say when Featurespace might go public.
“Why not be the next Autonomy?,” she said, when asked about the company’s ambitions. “We’ve got the experience and the team to get there.”

It has also attracted £1.7m from the Aim-listed Imperial Innovations, which has invested £161m since its 2006 IPO. Imperial Innovations now has a 22.1 per cent fully diluted equity stake in Featurespace.