ClimbsRocks/Concordia
Tracking machine learning: reconciling predictions at training time and serving time, tracking feature inputs and discrepancies over time, and more
PythonMIT
Issues
- 1
add option to only save top_n_features by feature importance, to save db size
#17 opened by ClimbsRocks - 0
MVP Scope
#8 opened by ClimbsRocks - 0
Add row_id_field per model
#16 opened by ClimbsRocks - 1
Interface for analytics
#15 opened by ClimbsRocks - 0
Goals 1/18
#14 opened by ClimbsRocks - 0
- 0
- 0
Supported use cases
#11 opened by ClimbsRocks - 0
FUTURE: delete_concordia
#10 opened by ClimbsRocks - 0
feature: let people add training rows multiple times for the same row_id, but for different model_ids
#9 opened by ClimbsRocks - 0
UI decisions
#6 opened by ClimbsRocks - 0
Scoping Decisions
#7 opened by ClimbsRocks - 0
do we want a prediction_id field?
#5 opened by ClimbsRocks - 0
Get getting predictions up and running
#4 opened by ClimbsRocks - 0
All DB operations should go through self.insert and self.retrieve, so that they can be overwritten by custom db logic
#3 opened by ClimbsRocks - 0
Initial Implementation Thoughts
#2 opened by ClimbsRocks - 0
Initial High Level Goals
#1 opened by ClimbsRocks