v0.5.0

DecisionFocusedLearningBenchmarks v0.5.0

Diff since v0.4.1

Breaking changes

Most of the benchmark interface has been reworked, it's now more generic and easy to use, but a good part of the API has changed (see the new documentation). All previous benchmarks are still there.

Merged pull requests:

• DataSample interface redesign (#52) (@BatyLeo)
• Enhance stochastic interface (#53) (@BatyLeo)
• Bump julia-actions/cache from 2 to 3 (#54) (@dependabot[bot])
• Global cleanup (#55) (@BatyLeo)
• Transform StochasticVehicleSchedulingBenchmark into a proper AbstractStochasticBenchmark{true} (#56) (@BatyLeo)
• Create aliases for exogenous/endogenous stochastic/dynamic benchmarks (#57) (@BatyLeo)
• Refactor visualization interface: optional Plots, uniform API (#58) (@BatyLeo)
• Support for contextual stochastic optimization problems (#60) (@BatyLeo)
• Cleanup and reinforce the interface (#62) (@BatyLeo)
• Switch to horizontal type hierarchy (#63) (@BatyLeo)

Closed issues:

• Move plotting dependencies into extensions (#29)
• Implement Plots.jl recipes instead of plotting interface functions (#33)
• Transform the stochastic vehicle scheduling benchmark into an AbstractStochasticBenchmark{true} (#34)
• Compatibility with TensorboardLogger? (#41)
• generate_policies should return a NamedTuple instead of a Tuple (#47)