toy_machines¶
from relational_datasets import load
train, test = load("toy_machines", "v0.0.6")
using RelationalDatasets
train, test = load("toy_machines", "v0.0.6")
This is a toy dataset based on one distributed with the ACE system (a short overview is provided on page 11 of "The ACE Data Mining System: User's Manual").
Machines contain parts, and some of those parts are either replaceable or
irreplaceable. Your goal is to infer whether the machine should be repaired
(replace), sent back to the manufacturer for repairs (sendback), or if the
machine is okay as it is (ok).
Task¶
Multiclass Classification: Is the machine ok, or should it be sendback
or replace?
machine(+id,#action).
worn(+id,-part).
replaceable(+part).
irreplaceable(+part).
range: part={gear, wheel, chain, engine, control_unit}.
range: action={ok, fix, sendback}.
Notes¶
A one-versus-rest classification scheme is recommended.
Last update:
November 9, 2022