Question



Clarification on Final Code Deliverable

Dr. Ballings,

You ask us this:

---------------------------------------------------------------------------------------------------------

#This is what the grader will use to evaluate your code:
#He will do this for multiple days and take the median AUC
#He will first load your functions. Then he will do the following:
setwd("path to all days")
model <- modelbuilding() #output is a model
symbolANDpred <- prediction(model) #last day in data is day to predict
#symbolANDpred is a dataframe with
two columns: symbol and prediction
#grader will make sure that his y's and your predictions are
accurately merged before calling the following line of code
auc(roc(symbolANDpred$prediction,y_lastday+1))

------------------------------------------------------------------------------------------------------------

If we use an ensemble method, then it will be multiple models.

We structured our code so that predictions and tests are done at the same time.

The models are stored in a list, do you want these instead of predictions?

If so, will the grader set up a loop to use the ensemble of them and average those for the final prediction?

Thanks,
Michael





Answers and follow-up questions





Answer or follow-up question 1

Dear Michael,

The modelbuilding() function can include multiple models, and other information.

The function for prediction can only return symbols and predictions.

Nothing can live outside of your functions and the grader will only use the code pasted in your question.

Michel Ballings



Sign in to be able to add an answer or mark this question as resolved.