cleanup + fix tests

Author: BatyLeo

src/DecisionFocusedLearningBenchmarks.jl

@@ -62,7 +62,7 @@ using .Utils
 
 # Interface
 export AbstractBenchmark, AbstractStochasticBenchmark, AbstractDynamicBenchmark, DataSample
-export AbstractEnv, get_seed, is_terminated, observe, reset!, step!
+export AbstractEnvironment, get_seed, is_terminated, observe, reset!, step!
 
 export Policy, run_policy!
 

src/DynamicAssortment/DynamicAssortment.jl

@@ -72,7 +72,7 @@ function Utils.generate_maximizer(b::DynamicAssortmentBenchmark)
 end
 
 function Utils.generate_environment(
-    ::DynamicAssortmentBenchmark, instance::Instance, rng::AbstractRNG
+    ::DynamicAssortmentBenchmark, instance::Instance, rng::AbstractRNG; kwargs...
 )
     seed = rand(rng, 1:typemax(Int))
     return Environment(instance; seed)

src/DynamicAssortment/policies.jl

@@ -19,23 +19,3 @@ function greedy_policy(env::Environment)
     maximizer = generate_maximizer(env.instance.config)
     return maximizer(prices(env))
 end
-
-function run_policy(env::Environment, episodes::Int; first_seed=1, policy=expert_policy)
-    dataset = []
-    rev_global = Float64[]
-    for i in 1:episodes
-        rev_episode = 0.0
-        CommonRLInterface.reset!(env; seed=first_seed - 1 + i, reset_seed=true)
-        training_instances = []
-        while !CommonRLInterface.terminated(env)
-            S = policy(env)
-            features = CommonRLInterface.observe(env)
-            push!(training_instances, DataSample(; x=features, y_true=S))
-            reward = CommonRLInterface.act!(env, S)
-            rev_episode += reward
-        end
-        push!(rev_global, rev_episode)
-        push!(dataset, training_instances)
-    end
-    return mean(rev_global), rev_global, dataset
-end

src/DynamicVehicleScheduling/DynamicVehicleScheduling.jl

@@ -25,27 +25,17 @@ include("static_vsp/parsing.jl")
 include("static_vsp/solution.jl")
 include("static_vsp/plot.jl")
 
-# dynamic environment
-include("environment/instance.jl")
-include("environment/state.jl")
-include("environment/scenario.jl")
-include("environment/environment.jl")
-include("environment/plot.jl")
+include("instance.jl")
+include("state.jl")
+include("scenario.jl")
+include("environment.jl")
+include("plot.jl")
 
-include("algorithms/prize_collecting_vsp.jl")
-include("algorithms/anticipative_solver.jl")
-
-include("learning/features.jl")
-include("learning/2d_features.jl")
+include("maximizer.jl")
+include("anticipative_solver.jl")
 
+include("features.jl")
 include("policy.jl")
-# include("policy/abstract_vsp_policy.jl")
-# include("policy/greedy_policy.jl")
-# include("policy/lazy_policy.jl")
-# include("policy/anticipative_policy.jl")
-# include("policy/kleopatra_policy.jl")
-
-include("maximizer.jl")
 
 """
 $TYPEDEF
@@ -84,7 +74,7 @@ function Utils.generate_dataset(b::DynamicVehicleSchedulingBenchmark, dataset_si
 end
 
 function Utils.generate_environment(
-    ::DynamicVehicleSchedulingBenchmark, instance::Instance, rng::AbstractRNG
+    ::DynamicVehicleSchedulingBenchmark, instance::Instance, rng::AbstractRNG; kwargs...
 )
     seed = rand(rng, 1:typemax(Int))
     return DVSPEnv(instance; seed)

src/DynamicVehicleScheduling/algorithms/prize_collecting_vsp.jl

@@ -40,3 +40,20 @@ end
 function compute_features(env::DVSPEnv)
     return compute_features(env.state, env.instance)
 end
+
+function get_features_meanTimeToRequests(state::DVSPState, instance::Instance)
+    quantiles = [0.5]
+    a = instance.static_instance.duration[state.location_indices, 2:end]
+    quantileTimeToRequests = mapslices(x -> quantile(x, quantiles), a; dims=2)
+    return quantileTimeToRequests
+end
+
+function compute_2D_features(state::DVSPState, instance::Instance)
+    timeDepotRequest = state.state_instance.duration[:, 1][state.is_postponable]
+    quantileTimeToRequests = get_features_meanTimeToRequests(state, instance)[state.is_postponable]
+    return hcat(timeDepotRequest, quantileTimeToRequests)'
+end
+
+function compute_2D_features(env::DVSPEnv)
+    return compute_2D_features(env.state, env.instance)
+end