Please enable JavaScript.
Coggle requires JavaScript to display documents.
Ch. 15: Build Candidate Models ("Importance Bar" (importance…
Ch. 15: Build Candidate Models
once you choose target..
you will see a True or False
if T and F are not well distributed, DataRobot will downsample
Under advanced options:
can choose different measures
for regression problems, other measures will be produced
15.2: save for later
Starting the analytical process:
will prepare data through the prescribed options: autopilot, quick, manual
quick run is abbreviated version of autopilot
produces almost just as good results
STEPS:
Setting target feature
Creating CV and Holdout partitions
Characterizing target variable
Loading dataset and preparing data
Saving target and partitioning information
Analyzing features
Calculating list of models
Informative Features represents all your features except for the ones automatically excluded and tagged, for example, as [Duplicate] or [Too Few Values]
"Importance Bar"
importance when examined against target independently of other features
kind of like an R squared indicator
*
hospital green bars not great, but perhaps the features will combine in great ways?