You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
read in each validation file.
2a. create a list of lists (predictionsAllRows)
create a list for each row (predictionsForRow)
append each algo's prediction to predictionsForRow
once we have read in all the predictions
read in the validation dataset (with all the features)
append predictionsAllRows to our validationData using hstack
post-MVP:
break out the data at this point into validationTrain and validationTest.
test is just going to be our actual test data.
from there, run a RF over the data.
post-post-post MVP:
run a modified version of machineJS over the data. see how we can train the best classifiers possible over the validation data and the predictions from the first round of machineJS.
then ensembler will simply average the results together.
what this means:
leave the current ensembler flow untouched. we will use that again on the second round, once we have run things back through machineJS.
we need to create a new workflow in machineJS to accommodate this (no splitData or dataFormatting. that might be the only difference)
focus:
within the validations folder:
2a. create a list of lists (predictionsAllRows)
read in the validation dataset (with all the features)
post-MVP:
break out the data at this point into validationTrain and validationTest.
test is just going to be our actual test data.
from there, run a RF over the data.
post-post-post MVP:
run a modified version of machineJS over the data. see how we can train the best classifiers possible over the validation data and the predictions from the first round of machineJS.
then ensembler will simply average the results together.
what this means:
leave the current ensembler flow untouched. we will use that again on the second round, once we have run things back through machineJS.
we need to create a new workflow in machineJS to accommodate this (no splitData or dataFormatting. that might be the only difference)
http://docs.scipy.org/doc/scipy-0.14.0/reference/generated/scipy.sparse.hstack.html
The text was updated successfully, but these errors were encountered: