Will the training time(speed) of our model be considered? I don’t see it in the grading criteria, but it’s mentioned in the title of the competition
I think the title of the competition is just based on this general idea: “Your model needs to learn from the data and be able to predict the price of an equity structured product faster than traditional computationally intensive models.” (from Alphien Dashboard)
and in terms of speed, only thing I’ve seen is “For this competition, your model shouldn’t take more than 2 hours to train.” (from Alphien Dashboard)
Otherwise, seems like model performance and explainability are the grading criteria
Hi, Heyangh,
The training time of your model will not be considered. The word “speed” contained in the name of the notebook refers somehow to the goal of the competition which is to return the price of a structured product faster than the traditional pricing models (as mentioned by Danpak94 in his post)
Adding up to Philippe’s comment above, the models once trained need to be fast. It needs to be a machine learning model and the sponsor (“UBS” here) has required us to check that models are fast. For an order of magnitude we would expect a prediction to be calculated in 0.1 second.
I hope that helps clarify and thank you for the good question.
Lionel.
And ofcourse if you want to calculate the execution time, either do a stopwatch using the time package or AlphienLab > Notebook cells show the execution time already
Manas