Improvements on benchmarks.sliding_window_benchmarks for robustness; Bugfixes Hwang and CVFTS methods

pyFTS/benchmarks/benchmarks.py

@@ -216,7 +216,6 @@ def sliding_window_benchmarks(data, windowsize, train=0.8, **kwargs):
         nodes = kwargs.get("nodes", ['127.0.0.1'])
         cluster, http_server = cUtil.start_dispy_cluster(experiment_method, nodes)
 
-    experiments = 0
     jobs = []
 
     inc = __pop("inc", 0.1, kwargs)
@@ -242,8 +241,12 @@ def sliding_window_benchmarks(data, windowsize, train=0.8, **kwargs):
                     if not distributed:
                         if progress:
                             progressbar.update(1)
-                        job = experiment_method(deepcopy(model), None, train, test, **kwargs)
+                        try:
-                        synthesis_method(dataset, tag, job, conn)
+                            job = experiment_method(deepcopy(model), None, train, test, **kwargs)
+                            synthesis_method(dataset, tag, job, conn)
+                        except Exception as ex:
+                            print("Error evaluating model " + model.shortname)
+                            print(ex)
                     else:
                         job = cluster.submit(deepcopy(model), None, train, test, **kwargs)
                         jobs.append(job)
@@ -257,7 +260,6 @@ def sliding_window_benchmarks(data, windowsize, train=0.8, **kwargs):
                 for partition in partitions:
 
                     for partitioner in partitioners_methods:
-                        print(transformation, partition)
 
                         data_train_fs = partitioner(data=train, npart=partition, transformation=transformation)
 
@@ -276,8 +278,12 @@ def sliding_window_benchmarks(data, windowsize, train=0.8, **kwargs):
                     if not distributed:
                         if progress:
                             progressbar.update(1)
-                        job = experiment_method(deepcopy(model), deepcopy(partitioner), train, test, **kwargs)
+                        try:
-                        synthesis_method(dataset, tag, job, conn)
+                            job = experiment_method(deepcopy(model), deepcopy(partitioner), train, test, **kwargs)
+                            synthesis_method(dataset, tag, job, conn)
+                        except Exception as ex:
+                            print("Error evaluating model " + model.shortname + " " + str(partitioner))
+                            print(ex)
                     else:
                         job = cluster.submit(deepcopy(model), deepcopy(partitioner), train, test, **kwargs)
                         job.id = id  # associate an ID to identify jobs (if needed later)

pyFTS/models/hwang.py

@@ -17,19 +17,16 @@ class HighOrderFTS(fts.FTS):
         self.name = "Hwang High Order FTS"
         self.shortname = "Hwang"
         self.detail = "Hwang"
+        self.configure_lags(**kwargs)
+
+    def configure_lags(self, **kwargs):
+        if "order" in kwargs:
+            self.order = kwargs.get("order", 2)
+
+        self.max_lag = self.order
 
     def forecast(self, ndata, **kwargs):
 
-        if 'order' in kwargs:
-            self.order = kwargs.get('order',self.order)
-            self.max_lag = self.order
-
-        if self.sets == None:
-            self.sets = self.partitioner.sets
-            ordered_sets = self.partitioner.ordered_sets
-        else:
-            ordered_sets = FuzzySet.set_ordered(self.sets)
-
         l = len(self.sets)
 
         cn = np.array([0.0 for k in range(l)])
@@ -42,17 +39,17 @@ class HighOrderFTS(fts.FTS):
         for t in np.arange(self.order-1, len(ndata)):
 
             for ix in range(l):
-                s = ordered_sets[ix]
+                s = self.partitioner.ordered_sets[ix]
-                cn[ix] = self.sets[s].membership( FuzzySet.grant_bounds(ndata[t], self.sets, ordered_sets))
+                cn[ix] = self.sets[s].membership( FuzzySet.grant_bounds(ndata[t], self.sets, self.partitioner.ordered_sets))
-                for w in range(self.order - 1):
+                for w in np.arange(self.order-1):
-                    ow[w, ix] = self.sets[s].membership(FuzzySet.grant_bounds(ndata[t - w], self.sets, ordered_sets))
+                    ow[w, ix] = self.sets[s].membership(FuzzySet.grant_bounds(ndata[t - w], self.sets, self.partitioner.ordered_sets))
                     rn[w, ix] = ow[w, ix] * cn[ix]
                     ft[ix] = max(ft[ix], rn[w, ix])
             mft = max(ft)
             out = 0.0
             count = 0.0
             for ix in range(l):
-                s = ordered_sets[ix]
+                s = self.partitioner.ordered_sets[ix]
                 if ft[ix] == mft:
                     out = out + self.sets[s].centroid
                     count += 1.0
@@ -61,4 +58,8 @@ class HighOrderFTS(fts.FTS):
         return ret
 
     def train(self, data, **kwargs):
-        pass
+
+        if self.sets == None:
+            self.sets = self.partitioner.sets
+
+        self.configure_lags(**kwargs)

pyFTS/models/nonstationary/cvfts.py

@@ -46,6 +46,7 @@ class ConditionalVarianceFTS(hofts.HighOrderFTS):
         self.uod_clip = False
         self.order = 1
         self.min_order = 1
+        self.max_lag = 1
         self.inputs = []
         self.forecasts = []
         self.residuals = []

pyFTS/tests/nonstationary.py

@@ -39,15 +39,44 @@ from pyFTS.partitioners import Grid, Util as pUtil
 from pyFTS.benchmarks import benchmarks as bchmk
 from pyFTS.models import chen
 
-tag = 'chen_partitioning'
+partitions = {'CMIV': {'BoxCox(0)': 17, 'Differential(1)': 7, 'None': 13},
+ 'IMCV': {'BoxCox(0)': 22, 'Differential(1)': 9, 'None': 25},
+ 'IMIV': {'BoxCox(0)': 27, 'Differential(1)': 11, 'None': 6},
+ 'NASDAQ': {'BoxCox(0)': 39, 'Differential(1)': 10, 'None': 34},
+ 'SP500': {'BoxCox(0)': 38, 'Differential(1)': 15, 'None': 39},
+ 'TAIEX': {'BoxCox(0)': 36, 'Differential(1)': 18, 'None': 38}}
 
-for ds in ['IMIV0']: #datasets.keys():
+
+tag = 'benchmarks'
+
+def nsfts_partitioner_builder(data, npart, transformation):
+    from pyFTS.partitioners import Grid
+    from pyFTS.models.nonstationary import perturbation, partitioners
+
+    tmp_fs = Grid.GridPartitioner(data=data, npart=npart, transformation=transformation)
+    fs = partitioners.SimpleNonStationaryPartitioner(data, tmp_fs,
+                                                     location=perturbation.polynomial,
+                                                     location_params=[1, 0],
+                                                     location_roots=0,
+                                                     width=perturbation.polynomial,
+                                                     width_params=[1, 0],
+                                                     width_roots=0)
+    return fs
+
+
+for ds in datasets.keys():
     dataset = datasets[ds]
 
-    bchmk.sliding_window_benchmarks(dataset, 4000, train=0.2, inc=0.2,
+    for tf in transformations.keys():
-                                    methods=[chen.ConventionalFTS],
+        transformation = transformations[tf]
-                                    benchmark_models=False,
+
-                                    transformations=[boxcox], #transformations[t] for t in transformations.keys()],
+        partitioning = partitions[ds][tf]
-                                    partitions=np.arange(3, 40, 1),
+
-                                    progress=False, type='point',
+        bchmk.sliding_window_benchmarks(dataset, 2000, train=0.2, inc=0.2,
-                                    file="nsfts_benchmarks.db", dataset=ds, tag=tag)
+                                        benchmark_models=False,
+                                        methods=[cvfts.ConditionalVarianceFTS],
+                                        partitioners_methods=[nsfts_partitioner_builder],
+                                        transformations=[transformation],
+                                        partitions=[partitioning],
+                                        progress=False, type='point',
+                                        file="nsfts_benchmarks.db", dataset=ds, tag=tag)