Merge branch 'dev'

Author: PasoStudio73

.github/workflows/ci.yml

@@ -2,7 +2,7 @@ name: CI
 on:
   push:
     branches:
-      - main
+      - dev
     tags: ['*']
   pull_request:
   workflow_dispatch:
@@ -24,9 +24,10 @@ jobs:
       matrix:
         version:
           - '1'
-          # - 'lts' broken: expects SolePostHoc to be registered
+          # - 'lts' # julia lts unsupported
         os:
           - ubuntu-latest
+          - ubuntu-24.04
         arch:
           - x64
     steps:
@@ -37,11 +38,11 @@ jobs:
           arch: ${{ matrix.arch }}
       - uses: julia-actions/cache@v2
       - uses: julia-actions/julia-buildpkg@v1
-        env:
-          PYTHON: ""
+        # env:
+        #   PYTHON: ""
       - uses: julia-actions/julia-runtest@v1
-        env:
-          PYTHON: ""
+        # env:
+        #   PYTHON: ""
       - uses: julia-actions/julia-processcoverage@v1
       - uses: codecov/codecov-action@v5
         with:

Project.toml

@@ -1,52 +1,51 @@
 name = "SoleXplorer"
 uuid = "5816ab97-8dbe-4a0a-b2fb-0334bd9876ed"
 version = "0.1.0"
-authors = ["Lorenzo BALBONI", "Federico MANZELLA", "Mauro MILELLA", "Giovanni PAGLIARINI", "Alberto PAPARELLA", "Riccardo PASINI", "Marco PERROTTA"]
+authors = ["Lorenzo BALBONI", "Michele GHIOTTI", "Federico MANZELLA", "Mauro MILELLA", "Giovanni PAGLIARINI", "Alberto PAPARELLA", "Riccardo PASINI", "Marco PERROTTA"]
 
 [deps]
-Catch22 = "acdeb78f-3d39-4310-8fdf-6d75c17c6d5a"
+CategoricalArrays = "324d7699-5711-5eae-9e2f-1d82baa6b597"
 DataFrames = "a93c6f00-e57d-5684-b7b6-d8193f3e46c0"
+DataTreatments = "1b3ff5f2-9b56-42a2-b38a-76a033f741ca"
+DimensionalData = "0703355e-b756-11e9-17c0-8b28908087d0"
+Imbalance = "c709b415-507b-45b7-9a3d-1767c89fde68"
 JLD2 = "033835bb-8acc-5ee8-8aae-3f567f8a3819"
 MLJ = "add582a8-e3ab-11e8-2d5e-e98b27df1bc7"
-MLJBalancing = "45f359ea-796d-4f51-95a5-deb1a414c586"
 MLJDecisionTreeInterface = "c6f25543-311c-4c74-83dc-3ea6d1015661"
 MLJParticleSwarmOptimization = "17a086e9-ed03-4f30-ab88-8b63f0f6126c"
 MLJXGBoostInterface = "54119dfa-1dab-4055-a167-80440f4f7a91"
-ModalAssociationRules = "fb95e5f7-b6b0-4744-a282-88ebeeae65a3"
 ModalDecisionTrees = "e54bda2e-c571-11ec-9d64-0242ac120002"
 Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
 Reexport = "189a3867-3050-52da-a836-e630ba90ab69"
-SoleBase = "4475fa32-7023-44a0-aa70-4813b230e492"
 SoleData = "123f1ae1-6307-4526-ab5b-aab3a92a2b8c"
 SoleModels = "4249d9c7-3290-4ddd-961c-e1d3ec2467f8"
 SolePostHoc = "cf1aa0c3-12c9-4ebe-9bdc-bd6c2ca79b72"
 XGBoost = "009559a3-9522-5dbb-924b-0b6ed2b22bb9"
 
 [compat]
-Catch22 = "0.7"
+CategoricalArrays = "0.10 - 1"
 DataFrames = "1"
+DataTreatments = "0.2"
+DimensionalData = "0.29"
+Imbalance = "0.1"
 JLD2 = "0.6"
 MLJ = "0.20, 0.21, 0.22"
 MLJBalancing = "0.1"
 MLJDecisionTreeInterface = "0.4"
 MLJParticleSwarmOptimization = "0.1"
 MLJXGBoostInterface = "0.3"
-ModalAssociationRules = "0.2"
 ModalDecisionTrees = "0.5"
 Random = "1"
 Reexport = "1"
-SoleBase = "0.13"
 SoleData = "0.16"
 SoleModels = "0.10"
 SolePostHoc = "0.1"
 XGBoost = "2.5"
 julia = "1"
 
 [extras]
-DelimitedFiles = "8bb1440f-4735-579b-a4ab-409b98df4dab"
-Downloads = "f43a241f-c20a-4ad4-852c-f6b1247861c6"
 JLD2 = "033835bb-8acc-5ee8-8aae-3f567f8a3819"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 
 [targets]
-test = ["DelimitedFiles", "Downloads", "JLD2", "Test"]
+test = ["JLD2", "Test"]

README.md

@@ -62,7 +62,7 @@ modelc = symbolic_analysis(
 
 ```julia
 # Load a temporal dataset
-natopsloader = NatopsLoader()
+natopsloader = SoleXplorer.NatopsLoader()
 Xts, yts = SoleXplorer.load(natopsloader)
 
 # Train a modal decision tree

TODO.md

@@ -13,8 +13,8 @@ resultsts = symbolic_analysis(
     Xts, yts;
     model=XGBoostClassifier(),
     resampling=(type=TimeSeriesCV(nfolds=5), rng=Xoshiro(1)),
-    win=AdaptiveWindow(nwindows=3, relative_overlap=0.3),
-    modalreduce=mean,
+    win=adaptivewindow(nwindows=3, overlap=0.3),
+    reducefunc=mean,
     features=[maximum, minimum],
     measures=(accuracy, log_loss, confusion_matrix, kappa)
 )
@@ -66,13 +66,6 @@ rivederne la propagazione: ad oggi non è ancora formalmente testata
 # ---------------------------------------------------------------------------- #
 Bisogna pensare anche ai modelli non supervisionati: apply(m, X)
 
-# ---------------------------------------------------------------------------- #
-#                                   timeout                                    #
-# ---------------------------------------------------------------------------- #
-l'algoritmo è pronto, sia il mio che quello mio di perry
-il più performante vince
-ma lo sviluppo sarà su un altra repo.
-
 # ---------------------------------------------------------------------------- #
 #                                multi thread                                  #
 # ---------------------------------------------------------------------------- #

docs/src/index.md

@@ -15,8 +15,8 @@ Key features:
 * Works also on time-series based datasets using modal logic.
 * View rules and their metrics.
 * Inspect logical formulas and their evaluation.
-* View modal rule associations.
-* Integrated GUI.
+<!-- * View modal rule associations. -->
+<!-- * Integrated GUI. -->
 
 ## Installation
 ```julia
@@ -59,7 +59,7 @@ modelc = symbolic_analysis(
 ### Temporal association rules
 ```julia
 # Load a temporal dataset
-natopsloader = NatopsLoader()
+natopsloader = SoleXplorer.NatopsLoader()
 Xts, yts = SoleXplorer.load(natopsloader)
 
 # Train a modal decision tree
@@ -78,7 +78,7 @@ SoleXplorer extensively uses the following packages:
 * [`SoleLogics`](https://github.com/aclai-lab/SoleLogics.jl): modal and temporal logic systems.
 * [`MLJ`](https://github.com/JuliaAI/MLJ.jl): provides all machine learning frameworks.
 * [`SolePostHoc`](https://github.com/aclai-lab/SolePostHoc.jl): for rule extraction.
-* [`ModalAssociationRules`](https://github.com/aclai-lab/ModalAssociationRules.jl): for mining association rules.
+<!-- * [`ModalAssociationRules`](https://github.com/aclai-lab/ModalAssociationRules.jl): for mining association rules. -->
 
 ## About
 The package is developed by the [ACLAI Lab](https://aclai.unife.it/en/) @ University of Ferrara.

docs/src/symbolic_analysis.md

@@ -20,7 +20,7 @@ ModelSet
 dsetup(m::ModelSet)
 solemodels(m::ModelSet)
 rules(m::ModelSet)
-associations(m::ModelSet)
+<!-- associations(m::ModelSet) -->
 performance(m::ModelSet)
 ```
 

docs/src/treatement.md

@@ -18,30 +18,30 @@ With multidimensional datasets there are two possible types of work:
 treatment(X::AbstractDataFrame, treat::Symbol)
 ```
 
-Windowing strategies availables for reduce/aggregation time-series datasets.
+<!-- Windowing strategies availables for reduce/aggregation time-series datasets.
 
 ```@docs
 MovingWindow
 WholeWindow
 SplitWindow
-AdaptiveWindow
+adaptivewindow
 AbstractWinFunction
 WinFunction
-```
+``` -->
 
 # [Featuresets](@id featuresets)
 
 ### Basic Statistics
 Standard statistical measures: `maximum`, `minimum`, `mean`, `median`, `std`, `cov`
 
-### Catch22 Features
+<!-- ### Catch22 Features
 Canonical time-series characteristics covering:
 - Distribution properties and extreme events
 - Linear and nonlinear autocorrelation structures  
 - Forecasting performance and scaling properties
-- Symbolic dynamics and transition patterns
+- Symbolic dynamics and transition patterns -->
 
-### Predefined Feature Sets
+<!-- ### Predefined Feature Sets
 
 - [`base_set`](@ref): Minimal statistical features (4 features)
 - [`catch9`](@ref): Curated subset combining statistics + key Catch22 (9 features)  
@@ -52,20 +52,20 @@ Canonical time-series characteristics covering:
 
 The Catch22 features are based on the Canonical Time-series Characteristics:
 - **Repository**: https://github.com/DynamicsAndNeuralSystems/catch22
-- **Paper**: Lubba, C.H., Sethi, S.S., Knaute, P. et al. "catch22: CAnonical Time-series CHaracteristics." *Data Min Knowl Disc* 33, 1821–1852 (2019). https://doi.org/10.1007/s10618-019-00647-x
+- **Paper**: Lubba, C.H., Sethi, S.S., Knaute, P. et al. "catch22: CAnonical Time-series CHaracteristics." *Data Min Knowl Disc* 33, 1821–1852 (2019). https://doi.org/10.1007/s10618-019-00647-x -->
 
-```@docs
+<!-- ```@docs
 base_set
 catch9
 catch22_set
 complete_set
-```
+``` -->
 
 See also: [`treatment`](@ref), [`setup_dataset`](@ref)
 
 ## All Catch22 Features
 
-```@docs
+<!-- ```@docs
 mode_5
 mode_10  
 embedding_dist
@@ -88,4 +88,4 @@ low_freq_power
 centroid_freq
 transition_variance
 periodicity
-```
+``` -->

src/SoleXplorer.jl

@@ -1,10 +1,10 @@
 module SoleXplorer
 using  Reexport
 
-using  SoleBase: Label, CLabel, RLabel, XGLabel
-using  SoleBase: movingwindow, wholewindow, splitwindow, adaptivewindow
 using  SoleData: scalarlogiset
 using  SoleData.Artifacts
+
+@reexport using SoleModels: Label, CLabel, RLabel, XGLabel
 using  SoleModels: Branch, ConstantModel
 using  SoleModels: DecisionEnsemble, DecisionTree, DecisionXGBoost
 using  SoleModels: AbstractModel, solemodel, weighted_aggregation, apply!
@@ -13,14 +13,15 @@ using  SoleModels: RuleExtractor, DecisionSet
 @reexport using SoleData.Artifacts: NatopsLoader, load
 @reexport using SoleModels: readmetrics
 
-@reexport using SolePostHoc: InTreesRuleExtractor, LumenRuleExtractor, BATreesRuleExtractor
-@reexport using SolePostHoc: RULECOSIPLUSRuleExtractor, REFNERuleExtractor, TREPANRuleExtractor
-using  SolePostHoc
+@reexport using SolePostHoc.RuleExtraction: InTreesRuleExtractor, LumenRuleExtractor, BATreesRuleExtractor
+# @reexport using SolePostHoc.RuleExtraction: RULECOSIPLUSRuleExtractor
+@reexport using SolePostHoc.RuleExtraction: REFNERuleExtractor, TREPANRuleExtractor
+using  SolePostHoc.RuleExtraction
 
-@reexport using ModalAssociationRules: Item, Atom, ScalarCondition, VariableMin, VariableMax
-@reexport using ModalAssociationRules: IA_L, box, diamond
-@reexport using ModalAssociationRules: gsupport, gconfidence, glift, gconviction, gleverage
-using ModalAssociationRules
+# @reexport using ModalAssociationRules: Item, Atom, ScalarCondition, VariableMin, VariableMax
+# @reexport using ModalAssociationRules: IA_L, box, diamond
+# @reexport using ModalAssociationRules: gsupport, gconfidence, glift, gconviction, gleverage
+# using ModalAssociationRules
 
 # ---------------------------------------------------------------------------- #
 #                                     MLJ                                      #
@@ -36,28 +37,27 @@ using  MLJParticleSwarmOptimization
 const  PSO = MLJParticleSwarmOptimization
 using  MLJ
 using  MLJ: MLJBase, MLJTuning
+# custom resampling strategy
+import MLJ.MLJBase: train_test_pairs
 # balancing
-using MLJBalancing
-@reexport using MLJBalancing:
+using Imbalance
+@reexport using Imbalance.MLJ:
     BorderlineSMOTE1, ClusterUndersampler, ENNUndersampler, ROSE,
     RandomOversampler, RandomUndersampler, RandomWalkOversampler,
     SMOTE, SMOTEN, SMOTENC, TomekUndersampler
 
 # ---------------------------------------------------------------------------- #
 #                              external packages                               #
 # ---------------------------------------------------------------------------- #
-@reexport using SoleData: load_arff_dataset
+@reexport using DataTreatments: movingwindow, wholewindow, splitwindow, adaptivewindow
+@reexport using DataTreatments: zscore, sigmoid, pnorm, scale, minmax, center, unitpower, outliersuppress
+using  DataTreatments
+
+using  CategoricalArrays
 using  DataFrames
 using  Random
 using  JLD2
 
-# ---------------------------------------------------------------------------- #
-#                                   timeout                                    #
-# ---------------------------------------------------------------------------- #
-# using TimeOut
-
-const preprocess = "using SoleXplorer\n"
-
 # ---------------------------------------------------------------------------- #
 #                                 maybe types                                  #
 # ---------------------------------------------------------------------------- #
@@ -76,25 +76,20 @@ const MaybeNTuple = Maybe{NamedTuple}
 # ---------------------------------------------------------------------------- #
 # feature extraction via Catch22
 # export user friendly Catch22 nicknames
-export mode_5, mode_10, embedding_dist, acf_timescale, acf_first_min, ami2,
-       trev, outlier_timing_pos, outlier_timing_neg, whiten_timescale,
-       forecast_error, ami_timescale, high_fluctuation, stretch_decreasing,
-       stretch_high, entropy_pairs, rs_range, dfa, low_freq_power, centroid_freq,
-       transition_variance, periodicity, base_set, catch9, catch22_set, complete_set
-using  Catch22
-include("featureset.jl")
+@reexport using DataTreatments: mode_5, mode_10, embedding_dist, acf_timescale,
+        acf_first_min, ami2, trev, outlier_timing_pos, outlier_timing_neg,
+        whiten_timescale, forecast_error, ami_timescale, high_fluctuation,
+        stretch_decreasing, stretch_high, entropy_pairs, rs_range, dfa,
+        low_freq_power, centroid_freq, transition_variance, periodicity, base_set
+@reexport using DataTreatments: catch9, catch22_set, complete_set
 
 # ---------------------------------------------------------------------------- #
 #                                 interfaces                                   #
 # ---------------------------------------------------------------------------- #
+export partition, pCV
 export get_X, get_y, get_train, get_test
 include("partition.jl")
 
-export AbstractWinFunction, WinFunction
-export MovingWindow, WholeWindow, SplitWindow, AdaptiveWindow
-export treatment
-include("treatment.jl")
-
 # ---------------------------------------------------------------------------- #
 #                                   models                                     #
 # ---------------------------------------------------------------------------- #
@@ -137,8 +132,8 @@ include("train_test.jl")
 
 include("extractrules.jl")
 
-export Apriori, FPGrowth, Eclat
-include("associationrules.jl")
+# export Apriori, FPGrowth, Eclat
+# include("associationrules.jl")
 
 export AbstractModelSet, ModelSet
 export dsetup, solemodels, rules, associations

src/apply.jl

@@ -72,7 +72,7 @@ get_classlabels(encoding)  = [string(encoding[i]) for i in sort(keys(encoding) |
 function apply(
     m :: Machine{DecisionTreeClassifier},
     X :: AbstractDataFrame,
-    y :: AbstractVector
+    y :: AbstractVector{<:CLabel}
 )::DecisionTree
     featurenames = MLJ.report(m).features
     classlabels  = sort(MLJ.report(m).classes_seen)