Betting the future on predictions: Kaggle

Local start-up Kaggle has developed a platform that has tackled some of the greatest dilemmas of our time: HIV prevention and predicting the winner of the Eurovision song contest.

Kaggle is a social-networking site where a community of statisticians compete to create models or methods to accurately predict a specific result or outcome.

It hosts competitions on behalf of companies or organisations which provide a limited set of data and a desired objective, whether this is predicting a future result or improving the method for results that have already been generated. A prize is awarded to the individual or team that most accurately predicts a result.

Past competitions have included developing models to predict how the HIV sequence evolves, forecasting global tourism numbers and competing against analysts from major financial institutions to predict the results of the 2010 soccer World Cup. It is currently running a competition to improve the world ranking system for chess players.

The data-prediction competition model was successfully used by United States company Netflix when it offered $1 million for the individual or team that could improve its movie recommendation service.

Kaggle's prize pool is modest by comparison, with hundreds of dollars and a variety of other prizes being handed out for five competitions.

The start-up was founded by Anthony Goldbloom, who previously performed macroeconomic modelling for the Reserve Bank of Australia, the Treasury and the ANZ Bank, and is chaired by the former head of the Govt 2.0 taskforce, Dr Nicholas Gruen.

Gruen said the company is not profitable, but explained that costs are low and expects as the business grows in the near future, revenues will be generated by a commission and consulting fees.

"The platform works because there are an infinite number of approaches that can be applied to any modelling task and it is impossible to know at the outset which technique will be most effective," he said.

"By opening a problem to a wide audience we very quickly get to the best that can be done given the inherent noise and richness of the dataset."

"We're so confident that we plan for most of our revenue to come from success fees."

Commentary

Strengths

Competitions are a proven way to connect statisticians with a company's forecasting requirements. Kaggle has a solid process to conduct competitions and there are some very talented professionals involved in the company.

Weaknesses

Like any social-networking site, Kaggle requires a large community to be successful. It faces a potential catch-22 scenario because it needs high-profile competitions to attract the best statisticians, and vice-versa.

Opportunities

There is a huge opportunity for this, moving into an era of where companies have access to huge amounts of data and information but there is a limited amount of resources to analyse this.

Also, because there is limited reliance on technology, the platform should scale quite easily.

Threats

The biggest threat will come from a competitor that can quickly develop an alternative platform, host high-profile competitions and widely market this capacity.

Gruen has said the focus is to build value for the community and clients while exploring other revenue models. If it can host some high-profile competition, to attract more users to the site, it will succeed.