- Refactoring: moving Ensemble methods to subpackage ensemble

- Parallel Seasonal Ensemble

ensemble/ensemble.py

@@ -12,6 +12,7 @@ from pyFTS.common import Transformations
 import scipy.stats as st
 from pyFTS import tree
 from pyFTS.models import msfts
+from pyFTS.probabilistic import ProbabilityDistribution, kde
 
 def sampler(data, quantiles):
     ret = []
@@ -242,28 +243,5 @@ class AllMethodEnsembleFTS(EnsembleFTS):
                     self.appendModel(model)
 
 
-class SeasonalEnsembleFTS(EnsembleFTS):
-    def __init__(self, name, **kwargs):
-        super(SeasonalEnsembleFTS, self).__init__(name="Seasonal Ensemble FTS", **kwargs)
-        self.min_order = 1
-        self.indexers = []
-        self.partitioners = []
-        self.is_multivariate = True
-        self.has_seasonality = True
-        self.has_probability_forecasting = True
-
-    def train(self, data, sets, order=1, parameters=None):
-        self.original_max = max(data)
-        self.original_min = min(data)
-
-        for ix in self.indexers:
-            for pt in self.partitioners:
-
-                model = msfts.MultiSeasonalFTS()
-                model.indexer = ix
-                model.appendTransformation(pt.transformation)
-                model.train(data,pt.sets,order=1)
-
-                self.appendModel(model)
 
 

ensemble/multiseasonal.py

@@ -0,0 +1,77 @@
+#!/usr/bin/python
+# -*- coding: utf8 -*-
+
+import numpy as np
+import pandas as pd
+import math
+from operator import itemgetter
+from pyFTS.common import FLR, FuzzySet, SortedCollection
+from pyFTS import fts, chen, cheng, hofts, hwang, ismailefendi, sadaei, song, yu, sfts
+from pyFTS.benchmarks import arima, quantreg
+from pyFTS.common import Transformations, Util as cUtil
+import scipy.stats as st
+from pyFTS.ensemble import ensemble
+from pyFTS.models import msfts
+from pyFTS.probabilistic import ProbabilityDistribution, kde
+from copy import deepcopy
+from joblib import Parallel, delayed
+import multiprocessing
+
+
+def train_individual_model(partitioner, train_data, indexer):
+    pttr = str(partitioner.__module__).split('.')[-1]
+    _key = "msfts_" + pttr + str(partitioner.partitions) + "_" + indexer.name
+
+    model = msfts.MultiSeasonalFTS(_key, indexer=indexer)
+    model.appendTransformation(partitioner.transformation)
+    model.train(train_data, partitioner.sets, order=1)
+
+    cUtil.persist_obj(model, "models/"+_key+".pkl")
+
+    print(_key)
+
+    return model
+
+
+class SeasonalEnsembleFTS(ensemble.EnsembleFTS):
+    def __init__(self, name, **kwargs):
+        super(SeasonalEnsembleFTS, self).__init__(name="Seasonal Ensemble FTS", **kwargs)
+        self.min_order = 1
+        self.indexers = []
+        self.partitioners = []
+        self.is_multivariate = True
+        self.has_seasonality = True
+        self.has_probability_forecasting = True
+
+    def train(self, data, sets, order=1, parameters=None):
+        self.original_max = max(data)
+        self.original_min = min(data)
+
+        num_cores = multiprocessing.cpu_count()
+
+        pool = {}
+        count = 0
+        for ix in self.indexers:
+            for pt in self.partitioners:
+                pool[count] = {'ix': ix, 'pt': pt}
+
+        results = Parallel(n_jobs=num_cores)(delayed(train_individual_model)(deepcopy(pool[m]['pt']), deepcopy(data), deepcopy(pool[m]['ix'])) for m in pool.keys())
+
+        for tmp in results:
+            self.appendModel(tmp)
+
+    def forecastDistribution(self, data, **kwargs):
+
+        ret = []
+
+        h = kwargs.get("h",10)
+
+        for k in data:
+
+            tmp = self.get_models_forecasts(k)
+
+            dist = ProbabilityDistribution.ProbabilityDistribution("KDE",h)
+
+            ret.append(dist)
+
+        return ret

models/msfts.py

@@ -7,7 +7,7 @@ class MultiSeasonalFTS(sfts.SeasonalFTS):
     """
     Multi-Seasonal Fuzzy Time Series
     """
-    def __init__(self, order, name, indexer, **kwargs):
+    def __init__(self, name, indexer, **kwargs):
         super(MultiSeasonalFTS, self).__init__("MSFTS")
         self.name = "Multi Seasonal FTS"
         self.shortname = "MSFTS " + name

models/seasonal/SeasonalIndexer.py

@@ -6,8 +6,9 @@ class SeasonalIndexer(object):
     """
     Seasonal Indexer. Responsible to find the seasonal index of a data point inside its data set
     """
-    def __init__(self,num_seasons):
+    def __init__(self,num_seasons, **kwargs):
         self.num_seasons = num_seasons
+        self.name = kwargs.get("name","")
 
     def get_season_of_data(self,data):
         pass
@@ -26,8 +27,8 @@ class SeasonalIndexer(object):
 
 
 class LinearSeasonalIndexer(SeasonalIndexer):
-    def __init__(self,seasons,units,ignore=None):
-        super(LinearSeasonalIndexer, self).__init__(len(seasons))
+    def __init__(self,seasons,units,ignore=None,**kwargs):
+        super(LinearSeasonalIndexer, self).__init__(len(seasons),kwargs)
         self.seasons = seasons
         self.units = units
         self.ignore = ignore
@@ -76,8 +77,8 @@ class LinearSeasonalIndexer(SeasonalIndexer):
 
 
 class DataFrameSeasonalIndexer(SeasonalIndexer):
-    def __init__(self,index_fields,index_seasons, data_fields):
-        super(DataFrameSeasonalIndexer, self).__init__(len(index_seasons))
+    def __init__(self,index_fields,index_seasons, data_fields,**kwargs):
+        super(DataFrameSeasonalIndexer, self).__init__(len(index_seasons),kwargs)
         self.fields = index_fields
         self.seasons = index_seasons
         self.data_fields = data_fields
@@ -131,8 +132,8 @@ class DateTime(Enum):
 
 
 class DateTimeSeasonalIndexer(SeasonalIndexer):
-    def __init__(self,date_field, index_fields, index_seasons, data_fields):
-        super(DateTimeSeasonalIndexer, self).__init__(len(index_seasons))
+    def __init__(self,date_field, index_fields, index_seasons, data_fields,**kwargs):
+        super(DateTimeSeasonalIndexer, self).__init__(len(index_seasons), kwargs)
         self.fields = index_fields
         self.seasons = index_seasons
         self.data_fields = data_fields

probabilistic/ProbabilityDistribution.py

@@ -2,6 +2,7 @@ import numpy as np
 import pandas as pd
 import matplotlib.pyplot as plt
 from pyFTS.common import FuzzySet,SortedCollection
+from pyFTS.probabilistic import kde
 
 
 class ProbabilityDistribution(object):
@@ -11,14 +12,15 @@ class ProbabilityDistribution(object):
     If type is KDE the PDF is continuous
     """
     def __init__(self,type, **kwargs):
+        self.uod = kwargs.get("uod", None)
+
         if type is None:
             self.type = "KDE"
+            self.kde = kde.KernelSmoothing(kwargs.get("h", 1), kwargs.get("method", "epanechnikov"))
         else:
             self.type = type
         self.description = kwargs.get("description", None)
 
-        self.uod = kwargs.get("uod", None)
-
         if self.type == "histogram":
             self.nbins = kwargs.get("num_bins", None)
             self.bins = kwargs.get("bins", None)
@@ -45,14 +47,15 @@ class ProbabilityDistribution(object):
             self.data.extend(values)
 
     def density(self, values):
-        if self.type == "histogram":
         ret = []
         for k in values:
+            if self.type == "histogram":
                 v = self.index.find_ge(k)
                 ret.append(self.distribution[v] / self.count)
-            return ret
             else:
-            pass
+                v = self.kde.probability(k, self.data)
+                ret.append(v)
+        return ret
 
 
     def cummulative(self, values):

tests/general.py

@@ -76,38 +76,33 @@ sonda_teste = sonda[1051201:]
 
 from pyFTS.models.seasonal import SeasonalIndexer
 
-ix_m15 = SeasonalIndexer.DateTimeSeasonalIndexer('data',[SeasonalIndexer.DateTime.minute],[15],'glo_avg')
+indexers = []
 
-cUtil.persist_obj(ix_m15, "models/sonda_ix_m15.pkl")
+for i in ["models/sonda_ix_m15.pkl", "models/sonda_ix_Mh.pkl", "models/sonda_ix_Mhm15.pkl"]:
+    obj = cUtil.load_obj(i)
+    indexers.append( obj )
+    print(obj)
+
+partitioners = []
+
+transformations = ["", "_diff"]
+for max_part in [10, 20, 30, 40, 50, 60]:
+    for t in transformations:
+        obj = cUtil.load_obj("models/sonda_fs_grid_" + str(max_part) + t + ".pkl")
+        partitioners.append( obj )
+        print(obj)
 
 
-ix_Mh = SeasonalIndexer.DateTimeSeasonalIndexer('data',[SeasonalIndexer.DateTime.month,SeasonalIndexer.DateTime.hour],
-                                             [None, None],'glo_avg')
+from pyFTS import ensemble
 
-cUtil.persist_obj(ix_Mh, "models/sonda_ix_Mh.pkl")
+fts = ensemble.SeasonalEnsembleFTS("")
 
-ix_Mhm15 = SeasonalIndexer.DateTimeSeasonalIndexer('data',[SeasonalIndexer.DateTime.month,
-                                                     SeasonalIndexer.DateTime.hour, SeasonalIndexer.DateTime.minute],
-                                             [None, None,15],'glo_avg')
+fts.indexers = indexers
+fts.partitioners = partitioners
 
-cUtil.persist_obj(ix_Mhm15, "models/sonda_ix_Mhm15.pkl")
-
-
-tmp = ix_Mh.get_data(sonda_treino)
-for max_part in [10, 20, 30, 40, 50]:
-
-    fs1 = Grid.GridPartitionerTrimf(tmp,max_part)
-
-    cUtil.persist_obj(fs1,"models/sonda_fs_grid_" + str(max_part) + ".pkl")
-
-    fs2 = FCM.FCMPartitionerTrimf(tmp, max_part)
-
-    cUtil.persist_obj(fs2, "models/sonda_fs_fcm_" + str(max_part) + ".pkl")
-
-    fs3 = Entropy.EntropyPartitionerTrimf(tmp, max_part)
-
-    cUtil.persist_obj(fs3, "models/sonda_fs_entropy_" + str(max_part) + ".pkl")
+fts.train(sonda_treino, sets=None)
 
+cUtil.persist_obj(fts, "models/msfts_ensemble_sonda_grid.pkl")
 
 from pyFTS.benchmarks import benchmarks as bchmk
 #from pyFTS.benchmarks import distributed_benchmarks as bchmk