The evaluation metric for this competition is Log Loss. The log loss score, commonly used in binary classification, measures the performance of the predicted probability given by a model. The log loss score is given by:
\n\\[ -{(y\\log(p) + (1 - y)\\log(1 - p))} \\]
\nLogarithmic loss measures the performance of a classification model where the prediction input is a probability value between 0 and 1. The goal of our machine learning models is to minimize this value. A perfect model would have a log loss of 0. Log loss increases as the predicted probability diverges from the actual label. So predicting a probability of .012 when the actual observation label is 1 would result in a high log loss.
\nThe file should contain a header and have the following format:
\nid,glasses\n4501,0.64\n4502,0.64\n4503,0.64\n4504,0.64\n4505,0.64\n4506,0.64\n4507,0.64\n4508,0.64\n...\n\nSome of the data in both the training and test ARE mislabeled. This is due to the algorithm that I used to generate the data. MOST of the data are labeled correctly, but there is some \"noise\". It may be possable to predict these mislabeled values, I am not sure.\n\n