Predictive analytics sees Springboks victory

Read time 4min 20sec
Local data scientists at Principa foresee a tightly-contested quarter final match between the Springboks and Wales.
Local data scientists at Principa foresee a tightly-contested quarter final match between the Springboks and Wales.

Using predictive analytics, local data scientists at Principa have foreseen a tightly-contested Rugby World Cup quarter final match between South Africa's Springboks and Wales tomorrow.

As the two teams lock horns at Twickenham Stadium in London, Principa predicts a four-point win in favour of the Springboks. SA have been one of the top try-scoring teams at this year's tournament, while Wales have hardly conceded any.

Jaco Rossouw, CEO of Principa, said it would take a lot more data to predict the actual score.

"We would need to look at what influences the final score a team gets, and the list of factors is probably endless."

Things like average score in the past, where the game is played, whether it is a pool game or knock-out game, who the referee is, and even the weather could all be inputs into the final score model, he explains.

"At this stage, we are finding predicting the margin challenging enough, but as a beta version, we could investigate how to take the model to the next level."

To come up with the prediction, the data scientists first took data from 6 000 matches played by 99 teams since 1995. Then as they progressed through the World Cup matches, the data scientists incrementally added the new data to their model from the matches played, so the model could adjust its expectations based on what happened in the last few matches.

"What's interesting is that the model doesn't look at who played the matches, but rather the characteristics of the teams that played the matches," says Rossouw.

"When we add that new data, the model adjusts and compares the characteristics of the teams whose data was added and the result of these teams' matches and adjusts its expectations for teams with the same or similar characteristics. We are continually adding new data and the model is continually adjusting to this new data, and as a result, our predictions become better with every match played," he explains.

It will take a lot more data to predict the actual score, says Jaco Rossouw, CEO of Principa.
It will take a lot more data to predict the actual score, says Jaco Rossouw, CEO of Principa.

Principa has a 93% "win success" in its Rugby World Cup predictions on sport prediction site, "We've gotten four predictions wrong, two of which were when Japan won, so we're not too happy with the Japanese team at the moment, since they're messing with our model's success rate," says Rossouw.

Describing the challenges data scientists face to make such predictions, Rossouw points to gathering the data, which is difficult and time-consuming.

"Once we have all of the data, it's just a matter of adding it to our model. Also getting dynamic data, not just historic, into the model that will change as we progress through the World Cup. For example, fantasy data, recent performance in the last match or last three matches, and the odds which change on a daily basis."

Tweet activity

Meanwhile, Twitter has been monitoring all the players taking part in #RWC2015 and have shared some of the most-tweeted moments, most-tweeted matches and the stats on the most-tweeted players and coaches.

Most-tweeted moments
1. The moment Japan seal a shock 34-32 victory against SA.
2. Full time whistle blows as England lose 25-28 to Wales in Pool A.
3. Japan score their final try to take a 34-32 lead over SA.
4. England become the first host nation to exit in group stages as they lose 13-33 to Australia.
5. The full time whistle sees Ireland run out 50-7 victors over Canada.
6. Scotland's Tommy Seymour tackled by Japan's Ayumu Goromaru just short of try line.
7. Japan's Akihito Yamada goes over the try line on 40+1 to make it Samoa 0-18 Japan.
8. Owen Farrell converts on full-time to seal England's 35-11 victory in the tournament opener.
9. Dan Biggar gives Wales a 28-25 lead over England with a 75th minute penalty.
10. England v Fiji kicks off to commence #RWC2015 at Twickenham.

Most-tweeted matches
1. England 13-33 Australia
2. England 35-11 Fiji
3. England 25-28 Wales
4. France 9-24 Ireland
5. South Africa 32-34 Japan
6. New Zealand 26-16 Argentina
7. Scotland 45-10 Japan
8. Australia 15-6 Wales
9 Ireland 50-7 Canada
10. Samoa 5-26 Japan

Most-tweeted players/coaches
1. Bryan Habana (@BryanHabana) - #RSA
2. Ayumu Goromaru - #JAP
3. Stuart Lancaster - #ENG
4. Dan Biggar - #WAL
5. Sam Burgess (@SamBurgess8) - #ENG
6. Mike Brown (@mikebrown_15) - #ENG
7. Jean de Villiers - #RSA
8. Owen Farrell (@owen_faz) -#ENG
9. Richie McCaw - #NZL
10. Sonny Bill Williams (@SonnyBWilliams) - #NZL

Login with