Submissions are evaluated on
[Multi Class Log Loss (MLogLoss)](https://github.com/jeffheaton/t81_558_deep_learning/blob/master/t81_558_class3_training.ipynb)
between the submitted outcome and the the observed outcome from the data set.
\[ \text{log loss} = -\frac{1}{N}\sum_{i=1}^N {( {y}_i\log(\hat{y}_i) + (1 - {y}_i)\log(1 - \hat{y}_i))} \]
Submission File
For each Id in the test set, you must predict the value of the outcome variable. The file should contain a header and have the following format:
Id,class-0,class-1,class-2,class-3,class-4
146,1,0,0,0,0
147,0,1,0,0,0
148,0.01,0.01,0.9,0.01,0.07
etc.