Please enable JavaScript.
Coggle requires JavaScript to display documents.
CHAPTERS 21-23 (SET UP PREDICTION SYSTEM: (Deployment Strategy (Drag-and…
CHAPTERS 21-23
SET UP PREDICTION SYSTEM:
model is given the data on that patient and will in turn provide a probability
probability is then turned into a suggested decision for the user of the the system
outside world changes
model success could be lower than expected
Deployment Strategy
Drag-and-drop
slow approach to scoring
Application Programming Interface (API)
allowing a programmer to write a program that uploads data for a new patient to this API and receive back a probability that the patient will be readmitted
DataRobot Prime
creates an approximation of your selected model available as code in the Python and Java programming languages
Batch
uses the DataRobot API to upload and score multiple large files in parallel
In-place with Spark
allows exporting the selected model as an executable file to be used in an Apache Spark (data processing) environment
DOCUMENT MODELING PROCCESS FOR REPRODUCABILITY :
important for others to understand what was done to accomplish the results and the justification for the project to exist
save future use of time on the project
document where the data came from
specify all the steps taken in DataRobot
record the business rules for use of the model and the probability thresholds
include info about how model was set up
CREATE MODEL MONITORING AND MAINTENANCE PLAN
tells others what to do in the event of changes in the environment that hurt the effectiveness of the model
format changes, variables change
DataRobot will fail rather than attempt to make the best of the available data
could result in target leakage
detect declining performance
evaluating the training data against new data
if model can distinguish 2 sets of data - indication that the business context has changed enough that the model should be retrained with more data