Competition Tasks and Metrics

In a competition, you start by examining the target metric. Understanding how your model’s errors are evaluated is key for scoring highly in every competition. When your predictions are submitted to the Kaggle platform, they are compared to a ground truth based on the target metric.

For instance, in the Titanic competition (https://www.kaggle.com/c/titanic/), all your submissions are evaluated based on accuracy, the percentage of surviving passengers you correctly predict. The organizers decided upon this metric because the aim of the competition is to find a model that estimates the probability of survival of a passenger under similar circumstances. In another knowledge competition, House Prices - Advanced Regression Techniques (https://www.kaggle.com/c/house-prices-advanced-regression-techniques), your work will be evaluated based on an average difference between your prediction and the ground truth. This involves computing the logarithm...