Machine Learning Canvas (v1.1)
Machine Learning Canvas (v1.1)
Type of task? Entity on which How are predictions turned Who is the end-user? What Strategy for initial train set & Where can we get (raw)
predictions are made? Possible into proposed value for the end-user? are their objectives? How will they continuous update. Mention collection information on entities and observed
outcomes? Wait time before Mention parameters of the process / benefit from the ML system? Mention rate, holdout on production entities, outcomes? Mention database tables,
observation? application that does that. workflow/interfaces. cost/constraints to observe outcomes. API methods, websites to scrape, etc.
Can models be deployed? When do we make real-time / How many prod models are Input representations
Which test data to assess performance? batch pred.? Time available for this + needed? When would we update? available at prediction time, extracted
Cost/gain values for (in)correct featurization + post-processing? Time available for this (including from raw data sources.
decisions? Fairness constraint? Compute target? featurization and analysis)?
MONITORING
Version 1.1. Created by Louis Dorard, Ph.D. Licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
Please keep this mention and the link to ownml.co when sharing.
OWNML.CO
ONLINE COURSE