An algorithm can predict if a film will be a hit, researchers claim   Leave a comment

People act unpredictably

In joint research, scholars from Oxford and Budapest claim to have devised an algorithm for predicting whether a film will score low or high at the box office. This would be music to the ears of producers, investors and studios. It sounds like nonsense to us, especially because the said model is entirely dependent on the historical buzz generated across Wikipedia, pontificating that other social media platforms are more susceptible to marketers’ manipulation. This segregation is contrived and unconvincing. To the limited extent that one media platform can be doctored, so can all. But the model’s central flaw is that it underestimates the uncertainty principle that underlies all human actions.

Of course it’s old hat for Hollywood to use focus groups to tweak a film’s final print to improve its odds of success. It’s also logical to tweak marketing campaigns to highlight those lines, characters or themes that are found to work in the first trailer. All this is not unlike psephology, where samples of an electorate are assumed to reflect the common opinion in an election. In all these cases, a percentage of present actions is used to estimate all of present actions. But it’s foolhardy to think that an algorithm predicated on the past can apply to the majority of the planet in the future. Or that a prophesied combination of kisses, car chases and bomb explosions can guarantee a blockbuster.

Not only are people’s tastes mercurial, the film market is fast changing. It feels like one day Bollywood was genuflecting to Rs 100 crore hits and the next day this membership fee doubled. All the certitudes have uncertain counterparts. All the Khans have suffered their fair share of flops. Sometimes statisticians’ hubris is without end. But Yogi Berra is still right: It’s tough to make predictions, especially about the future.


Don’t underestimate big data

Sanjiv Shankaran

According to IBM, 90% of data in the world has been created in just the last two years. We live in the age of big data, which keeps constantly growing as it spans everything from a text message sent over a mobile phone to data collection by census workers. Marry big data to ever improving techniques of statistical analyses and we enter the world of greater predictive powers which show up everywhere, from policing in the US to weather forecasts across the globe. As more and more people leave digital footprints, data scientists will begin to steer decision-making away from hunches into the realm of rationality.

Diffusion of technology has begun to generate enormous amount of data. Mere computing power is not enough to get the most out of it. Human intellect will remain the prized possession even in this era as greater the data, the more important it is to have a mental framework to separate the wheat from the chaff. But no matter which way one approaches the subject, it is the availability of adequate data that will act as a force multiplier to human intellect and intuition. A question that comes up frequently is whether statistical analysis can accurately predict the way human beings will react to different situations.

The story of Nate Silver, an American statistician, and the last two US presidential elections convincingly proves that collective behaviour is indeed predictable. The last two presidential elections were held in the backdrop of an extraordinary economic crisis as well as a transformation in the country’s demography. Not an easy environment to make a prediction. Yet, in both elections, not only did Silver correctly predict the overall verdict; he also got it right in the states. In the 2008 election, Silver got it right in 49 of the 50 states, and all 50 in the next election.


Posted August 28, 2013 by avinash2060 in Machine Learning

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: