The linear regression model used predicts the outcome of each match and the winning margin by targeting the points difference for each game and variables built around the form of each team coming into the match (win percent, tries per game etc), their ranking difference compared to the opponent, and their recent history against that particular opponent (last win/loss margin, win percent). With this model, Alteryx predicts the outcome of the pool matches and the route to World Cup triumph as:
Quarter-finals
· Ireland to beat South Africa by 7 points
· New Zealand to beat Scotland by 18 points
· England to beat Australia by 11 points
· Wales to beat France by 12 points
Semi-finals
· Wales to beat Ireland by 7 points
· New Zealand to beat England by 3 points
Final
· New Zealand to beat Wales by 8 points for a third consecutive title.
Alan Jacobson, Chief Data and Analytics Officer at Alteryx comments on these predictions; “Whether you need to crunch big data sets or tackle complex analytical predictions to increase percentages, data has become a vital part of both player and game appraisal analysis. From performance data accrued through wireless-enabled wearable activity trackers to predicting training loads or identifying key injury triggers, predictive and spatial tools are changing the way sports are played as teams turn to data to achieve a competitive edge. By building advanced predictive models to exploit past match and individual performance data in order to garner complex insights we can realistically predict New Zealand will be the ultimate tournament winner in Japan.”