k-Nearest Neighbors benchmark method

pyFTS/benchmarks/Measures.py

@@ -387,7 +387,7 @@ def get_distribution_statistics(data, model, **kwargs):
         _s1 = time.time()
         forecasts = model.predict(data, **kwargs)
         _e1 = time.time()
-        ret.append(round(crps(data, forecasts), 3))
+        ret.append(round(crps(data[model.order:], forecasts), 3))
         ret.append(round(_e1 - _s1, 3))
     else:
         skip = kwargs.get('steps_ahead_sampler', 1)

pyFTS/benchmarks/arima.py

@@ -30,7 +30,7 @@ class ARIMA(fts.FTS):
         self.benchmark_only = True
         self.min_order = 1
         self.alpha = kwargs.get("alpha", 0.05)
-        self.shortname += str(self.alpha)
+        self.order = kwargs.get("order", (1,0,0))
         self._decompose_order(self.order)
 
     def _decompose_order(self, order):
@@ -43,6 +43,10 @@ class ARIMA(fts.FTS):
             self.shortname = "ARIMA(" + str(self.p) + "," + str(self.d) + "," + str(self.q) + ") - " + str(self.alpha)
 
     def train(self, data, **kwargs):
+
+        self.original_min = np.nanmin(data)
+        self.original_max = np.nanmax(data)
+
         if kwargs.get('order', None) is not None:
             order = kwargs.get('order', (1,0,0))
             self._decompose_order(order)
@@ -50,8 +54,6 @@ class ARIMA(fts.FTS):
         if self.indexer is not None:
             data = self.indexer.get_data(data)
 
-        #data = self.apply_transformations(data, updateUoD=True)
-
         try:
             self.model =  stats_arima(data, order=(self.p, self.d, self.q))
             self.model_fit = self.model.fit(disp=0)
@@ -69,9 +71,6 @@ class ARIMA(fts.FTS):
         if self.model_fit is None:
             return np.nan
 
-        if self.indexer is not None and isinstance(ndata, pd.DataFrame):
-            data = self.indexer.get_data(ndata)
-
         ndata = np.array(ndata)
 
         l = len(ndata)
@@ -101,8 +100,6 @@ class ARIMA(fts.FTS):
 
         sigma = np.sqrt(self.model_fit.sigma2)
 
-        #ndata = np.array(self.apply_transformations(data))
-
         l = len(data)
 
         ret = []
@@ -122,8 +119,6 @@ class ARIMA(fts.FTS):
 
             ret.append(tmp)
 
-        #ret = self.apply_inverse_transformations(ret, params=[data[self.order - 1:]], point_to_interval=True)
-
         return ret
 
     def forecast_ahead_interval(self, ndata, steps, **kwargs):
@@ -134,8 +129,6 @@ class ARIMA(fts.FTS):
 
         sigma = np.sqrt(self.model_fit.sigma2)
 
-        #ndata = np.array(self.apply_transformations(data))
-
         l = len(ndata)
 
         nmeans = self.forecast_ahead(ndata, steps, **kwargs)
@@ -152,15 +145,10 @@ class ARIMA(fts.FTS):
 
             ret.append(tmp)
 
-        #ret = self.apply_inverse_transformations(ret, params=[[data[-1] for a in np.arange(0, steps)]], interval=True)
-
         return ret
 
     def forecast_distribution(self, data, **kwargs):
 
-        if self.indexer is not None and isinstance(data, pd.DataFrame):
-            data = self.indexer.get_data(data)
-
         sigma = np.sqrt(self.model_fit.sigma2)
 
         l = len(data)
@@ -168,8 +156,6 @@ class ARIMA(fts.FTS):
         ret = []
 
         for k in np.arange(self.order, l + 1):
-            tmp = []
-
             sample = [data[i] for i in np.arange(k - self.order, k)]
 
             mean = self.forecast(sample)

pyFTS/benchmarks/benchmarks.py

@@ -19,7 +19,7 @@ from pyFTS.probabilistic import ProbabilityDistribution
 from pyFTS.common import Transformations
 from pyFTS.models import song, chen, yu, ismailefendi, sadaei, hofts, pwfts, ifts, cheng, hwang
 from pyFTS.models.ensemble import ensemble
-from pyFTS.benchmarks import Measures, naive, arima, ResidualAnalysis, quantreg
+from pyFTS.benchmarks import Measures, naive, arima, ResidualAnalysis, quantreg, knn
 from pyFTS.benchmarks import Util as bUtil
 from pyFTS.common import Util as cUtil
 # from sklearn.cross_validation import KFold
@@ -156,8 +156,7 @@ def sliding_window_benchmarks(data, windowsize, train=0.8, **kwargs):
         elif benchmark_methods is not None:
             for count, model in enumerate(benchmark_methods, start=0):
                 par = benchmark_methods_parameters[count]
-                mfts = model(str(par if par is not None else ""))
+                mfts = model("", **par)
-                mfts.order = par
                 pool.append(mfts)
 
     if type == 'point':
@@ -244,7 +243,6 @@ def sliding_window_benchmarks(data, windowsize, train=0.8, **kwargs):
         progressbar.close()
 
     if distributed:
-        jobs2 = []
 
         rng = jobs
 
@@ -268,10 +266,6 @@ def sliding_window_benchmarks(data, windowsize, train=0.8, **kwargs):
 
     conn.close()
 
-    sintetic = kwargs.get('sintetic', False)
-
-    #return synthesis_method(jobs, experiments, save, file, sintetic)
-
 
 def get_benchmark_point_methods():
     """Return all non FTS methods for point forecasting"""
@@ -287,7 +281,7 @@ def get_point_methods():
 
 def get_benchmark_interval_methods():
     """Return all non FTS methods for point_to_interval forecasting"""
-    return [quantreg.QuantileRegression]
+    return [ arima.ARIMA, quantreg.QuantileRegression]
 
 
 def get_interval_methods():
@@ -302,7 +296,7 @@ def get_probabilistic_methods():
 
 def get_benchmark_probabilistic_methods():
     """Return all FTS methods for probabilistic forecasting"""
-    return [arima.ARIMA, quantreg.QuantileRegression]
+    return [arima.ARIMA, quantreg.QuantileRegression, knn.KNearestNeighbors]
 
 
 def run_point(mfts, partitioner, train_data, test_data, window_key=None, **kwargs):
@@ -398,6 +392,7 @@ def run_interval(mfts, partitioner, train_data, test_data, window_key=None, **kw
     method = kwargs.get('method', None)
 
     if mfts.benchmark_only:
+        mfts.append_transformation(partitioner.transformation)
         _key = mfts.shortname + str(mfts.order if mfts.order is not None else "") + str(mfts.alpha)
     else:
         pttr = str(partitioner.__module__).split('.')[-1]
@@ -444,10 +439,11 @@ def run_probabilistic(mfts, partitioner, train_data, test_data, window_key=None,
     from pyFTS.models import hofts, ifts, pwfts
     from pyFTS.models.ensemble import ensemble
     from pyFTS.partitioners import Grid, Entropy, FCM
-    from pyFTS.benchmarks import Measures, arima
+    from pyFTS.benchmarks import Measures, arima, quantreg, knn
     from pyFTS.models.seasonal import SeasonalIndexer
 
-    tmp = [hofts.HighOrderFTS, ifts.IntervalFTS, pwfts.ProbabilisticWeightedFTS, arima.ARIMA, ensemble.AllMethodEnsembleFTS]
+    tmp = [hofts.HighOrderFTS, ifts.IntervalFTS, pwfts.ProbabilisticWeightedFTS, arima.ARIMA,
+           ensemble.AllMethodEnsembleFTS, knn.KNearestNeighbors]
 
     tmp2 = [Grid.GridPartitioner, Entropy.EntropyPartitioner, FCM.FCMPartitioner]
 
@@ -460,6 +456,7 @@ def run_probabilistic(mfts, partitioner, train_data, test_data, window_key=None,
 
     if mfts.benchmark_only:
         _key = mfts.shortname + str(mfts.order if mfts.order is not None else "") + str(mfts.alpha)
+        mfts.append_transformation(partitioner.transformation)
     else:
         pttr = str(partitioner.__module__).split('.')[-1]
         _key = mfts.shortname + " n = " + str(mfts.order) + " " + pttr + " q = " + str(partitioner.partitions)

pyFTS/benchmarks/knn.py

@@ -0,0 +1,71 @@
+#!/usr/bin/python
+# -*- coding: utf8 -*-
+
+import numpy as np
+from statsmodels.tsa.tsatools import lagmat
+from pyFTS.common import fts
+from pyFTS.probabilistic import ProbabilityDistribution
+
+class KNearestNeighbors(fts.FTS):
+    """
+    K-Nearest Neighbors
+    """
+    def __init__(self, name, **kwargs):
+        super(KNearestNeighbors, self).__init__(1, "kNN"+name)
+        self.name = "kNN"
+        self.detail = "K-Nearest Neighbors"
+        self.is_high_order = True
+        self.has_point_forecasting = True
+        self.has_interval_forecasting = True
+        self.has_probability_forecasting = True
+        self.benchmark_only = True
+        self.min_order = 1
+        self.alpha = kwargs.get("alpha", 0.05)
+        self.order = kwargs.get("order", 1)
+        self.lag = None
+        self.k = kwargs.get("k", 30)
+
+    def train(self, data, **kwargs):
+        if kwargs.get('order', None) is not None:
+            self.order = kwargs.get('order', 1)
+
+        self.data = data
+
+        #self.lagdata, = lagmat(data, maxlag=self.order, trim="both", original='sep')
+
+
+    def knn(self, sample):
+
+        if self.order == 1:
+            dist = np.apply_along_axis(lambda x: (x - sample) ** 2, 0, self.data)
+            ix = np.argsort(dist) + 1
+        else:
+            dist = []
+            for k in np.arange(self.order, len(self.data)):
+                dist.append(sum([ (self.data[k - kk] - sample[kk])**2 for kk in range(self.order)]))
+            ix = np.argsort(np.array(dist)) + self.order + 1
+
+        ix = np.clip(ix, 0, len(self.data)-1 )
+        return self.data[ix[:self.k]]
+
+    def forecast_distribution(self, data, **kwargs):
+        ret = []
+
+        smooth = kwargs.get("smooth", "KDE")
+        alpha = kwargs.get("alpha", None)
+
+        uod = self.get_UoD()
+
+        for k in np.arange(self.order, len(data)):
+
+            sample = data[k-self.order : k]
+
+            forecasts = self.knn(sample)
+
+            dist = ProbabilityDistribution.ProbabilityDistribution(smooth, uod=uod, data=forecasts,
+                                                                   name="", **kwargs)
+            ret.append(dist)
+
+        return ret
+
+

pyFTS/benchmarks/quantreg.py

@@ -19,7 +19,7 @@ class QuantileRegression(fts.FTS):
         self.has_interval_forecasting = True
         self.has_probability_forecasting = True
         self.benchmark_only = True
-        self.minOrder = 1
+        self.min_order = 1
         self.alpha = kwargs.get("alpha", 0.05)
         self.dist = kwargs.get("dist", False)
         self.upper_qt = None
@@ -28,15 +28,14 @@ class QuantileRegression(fts.FTS):
         self.dist_qt = None
         self.shortname = "QAR("+str(self.order)+","+str(self.alpha)+")"
 
-    def train(self, data, sets, order=1, parameters=None):
+    def train(self, data, **kwargs):
-        self.order = order
+        if kwargs.get('order', None) is not None:
+            self.order = kwargs.get('order', 1)
 
         if self.indexer is not None and isinstance(data, pd.DataFrame):
             data = self.indexer.get_data(data)
 
-        tmp = np.array(self.apply_transformations(data, updateUoD=True))
+        lagdata, ndata = lagmat(data, maxlag=self.order, trim="both", original='sep')
-
-        lagdata, ndata = lagmat(tmp, maxlag=order, trim="both", original='sep')
 
         mqt = QuantReg(ndata, lagdata).fit(0.5)
         if self.alpha is not None:
@@ -76,12 +75,8 @@ class QuantileRegression(fts.FTS):
         up = self.linearmodel([k[1] for k in data], up_params)
         return [lo, up]
 
-    def forecast(self, data, **kwargs):
+    def forecast(self, ndata, **kwargs):
 
-        if self.indexer is not None and isinstance(data, pd.DataFrame):
-            data = self.indexer.get_data(data)
-
-        ndata = np.array(self.apply_transformations(data))
         l = len(ndata)
 
         ret = []
@@ -91,16 +86,9 @@ class QuantileRegression(fts.FTS):
 
             ret.append(self.linearmodel(sample, self.mean_qt))
 
-        ret = self.apply_inverse_transformations(ret, params=[data[self.order - 1:]])
-
         return ret
 
-    def forecast_interval(self, data, **kwargs):
+    def forecast_interval(self, ndata, **kwargs):
-
-        if self.indexer is not None and isinstance(data, pd.DataFrame):
-            data = self.indexer.get_data(data)
-
-        ndata = np.array(self.apply_transformations(data))
 
         l = len(ndata)
 
@@ -110,16 +98,9 @@ class QuantileRegression(fts.FTS):
             sample = ndata[k - self.order: k]
             ret.append(self.point_to_interval(sample, self.lower_qt, self.upper_qt))
 
-        ret = self.apply_inverse_transformations(ret, params=[data[self.order - 1:]], interval=True)
-
         return ret
 
-    def forecast_ahead_interval(self, data, steps, **kwargs):
+    def forecast_ahead_interval(self, ndata, steps, **kwargs):
-
-        if self.indexer is not None and isinstance(data, pd.DataFrame):
-            data = self.indexer.get_data(data)
-
-        ndata = np.array(self.apply_transformations(data))
 
         smoothing = kwargs.get("smoothing", 0.9)
 
@@ -137,20 +118,13 @@ class QuantileRegression(fts.FTS):
 
             ret.append([intl[0]*(1 + k*smoothing), intl[1]*(1 + k*smoothing)])
 
-        ret = self.apply_inverse_transformations(ret, params=[[data[-1] for a in np.arange(0, steps + self.order)]], interval=True)
-
         return ret[-steps:]
 
-    def forecast_distribution(self, data, **kwargs):
+    def forecast_distribution(self, ndata, **kwargs):
-
-        if self.indexer is not None and isinstance(data, pd.DataFrame):
-            data = self.indexer.get_data(data)
-
-        ndata = np.array(self.apply_transformations(data))
 
         ret = []
 
-        l = len(data)
+        l = len(ndata)
 
         for k in np.arange(self.order, l + 1):
             dist = ProbabilityDistribution.ProbabilityDistribution(type="histogram",
@@ -167,12 +141,7 @@ class QuantileRegression(fts.FTS):
 
         return ret
 
-    def forecast_ahead_distribution(self, data, steps, **kwargs):
+    def forecast_ahead_distribution(self, ndata, steps, **kwargs):
-
-        if self.indexer is not None and isinstance(data, pd.DataFrame):
-            data = self.indexer.get_data(data)
-
-        ndata = np.array(self.apply_transformations(data))
 
         ret = []
 

pyFTS/common/flrg.py

@@ -42,7 +42,6 @@ class FLRG(object):
     def get_membership(self, data, sets):
         ret = 0.0
         if isinstance(self.LHS, (list, set)):
-            assert len(self.LHS) == len(data)
             ret = np.nanmin([sets[self.LHS[ct]].membership(dat) for ct, dat in enumerate(data)])
         else:
             ret = sets[self.LHS].membership(data)

pyFTS/models/ensemble/ensemble.py

@@ -5,6 +5,7 @@ import numpy as np
 import pandas as pd
 from pyFTS.common import SortedCollection, fts, tree
 from pyFTS.models import chen, cheng, hofts, hwang, ismailefendi, sadaei, song, yu
+from pyFTS.probabilistic import ProbabilityDistribution
 import scipy.stats as st
 
 
@@ -171,29 +172,52 @@ class EnsembleFTS(fts.FTS):
 
         return ret
 
-    def empty_grid(self, resolution):
+    def forecast_distribution(self, data, **kwargs):
-        return self.get_empty_grid(-(self.original_max*2), self.original_max*2, resolution)
+        ret = []
+
+        smooth = kwargs.get("smooth", "KDE")
+        alpha = kwargs.get("alpha", None)
+
+        uod = self.get_UoD()
+
+        for k in np.arange(self.order, len(data)):
+
+            sample = data[k-self.order : k]
+
+            forecasts = self.get_models_forecasts(sample)
+
+            if alpha is None:
+                forecasts = np.ravel(forecasts).tolist()
+            else:
+                forecasts = self.get_distribution_interquantile(np.ravel(forecasts).tolist(), alpha)
+
+            dist = ProbabilityDistribution.ProbabilityDistribution(smooth, uod=uod, data=forecasts,
+                                                                   name="", **kwargs)
+
+            ret.append(dist)
+
+        return ret
+
 
     def forecast_ahead_distribution(self, data, steps, **kwargs):
         if 'method' in kwargs:
             self.point_method = kwargs.get('method','mean')
 
-        percentile_size = (self.original_max - self.original_min) / 100
+        smooth = kwargs.get("smooth", "KDE")
-
+        alpha = kwargs.get("alpha", None)
-        resolution = kwargs.get('resolution', percentile_size)
-
-        grid = self.empty_grid(resolution)
-
-        index = SortedCollection.SortedCollection(iterable=grid.keys())
 
         ret = []
 
-        samples = [[k] for k in data[-self.order:]]
+        start = kwargs.get('start', self.order)
 
-        for k in np.arange(self.order, steps + self.order):
+        uod = self.get_UoD()
+
+        sample = data[start - self.order: start]
+
+        for k in np.arange(self.order, steps+self.order):
             forecasts = []
             lags = {}
-            for i in np.arange(0, self.order): lags[i] = samples[k - self.order + i]
+            for i in np.arange(0, self.order): lags[i] = sample[k-self.order]
 
             # Build the tree with all possible paths
 
@@ -206,17 +230,19 @@ class EnsembleFTS(fts.FTS):
 
                 forecasts.extend(self.get_models_forecasts(path))
 
-            samples.append(sampler(forecasts, np.arange(0.1, 1, 0.1)))
+            sample.append(sampler(forecasts, np.arange(0.1, 1, 0.1)))
 
-            grid = self.gridCountPoint(grid, resolution, index, forecasts)
+            if alpha is None:
+                forecasts = np.ravel(forecasts).tolist()
+            else:
+                forecasts = self.get_distribution_interquantile(np.ravel(forecasts).tolist(), alpha)
 
-            tmp = np.array([grid[i] for i in sorted(grid)])
+            dist = ProbabilityDistribution.ProbabilityDistribution(smooth, uod=uod, data=forecasts,
+                                                                   name="", **kwargs)
 
-            ret.append(tmp / sum(tmp))
+            ret.append(dist)
 
-        grid = self.empty_grid(resolution)
+        return ret
-        df = pd.DataFrame(ret, columns=sorted(grid))
-        return df
 
 
 class AllMethodEnsembleFTS(EnsembleFTS):

pyFTS/models/ifts.py

@@ -26,6 +26,7 @@ class IntervalFTS(hofts.HighOrderFTS):
         self.has_point_forecasting = False
         self.has_interval_forecasting = True
         self.is_high_order = True
+        self.min_order = 1
 
     def get_upper(self, flrg):
         if flrg.get_key() in self.flrgs:

pyFTS/models/pwfts.py

@@ -105,6 +105,7 @@ class ProbabilisticWeightedFTS(ifts.IntervalFTS):
         self.has_interval_forecasting = True
         self.has_probability_forecasting = True
         self.is_high_order = True
+        self.min_order = 1
         self.auto_update = kwargs.get('update',False)
 
 

pyFTS/probabilistic/ProbabilityDistribution.py

@@ -14,15 +14,25 @@ class ProbabilityDistribution(object):
     def __init__(self, type = "KDE", **kwargs):
         self.uod = kwargs.get("uod", None)
 
-        self.type = type
+        self.data = []
-        if self.type == "KDE":
-            self.kde = kde.KernelSmoothing(kwargs.get("h", 0.5), kwargs.get("kernel", "epanechnikov"))
 
-        self.nbins = kwargs.get("num_bins", 100)
+        self.type = type
 
         self.bins = kwargs.get("bins", None)
         self.labels = kwargs.get("bins_labels", None)
 
+        data = kwargs.get("data", None)
+
+        if self.type == "KDE":
+            self.kde = kde.KernelSmoothing(kwargs.get("h", 0.5), kwargs.get("kernel", "epanechnikov"))
+            _min = np.nanmin(data)
+            _min = _min * .7 if _min > 0 else _min * 1.3
+            _max = np.nanmax(data)
+            _max = _max * 1.3 if _max > 0 else _max * .7
+            self.uod = [_min, _max]
+
+        self.nbins = kwargs.get("num_bins", 100)
+
         if self.bins is None:
             self.bins = np.linspace(int(self.uod[0]), int(self.uod[1]), int(self.nbins)).tolist()
             self.labels = [str(k) for k in self.bins]
@@ -38,10 +48,6 @@ class ProbabilityDistribution(object):
         self.count = 0
         for k in self.bins: self.distribution[k] = 0
 
-        self.data = []
-
-        data = kwargs.get("data",None)
-
         if data is not None:
             self.append(data)
 
@@ -228,10 +234,12 @@ class ProbabilityDistribution(object):
 
     def __str__(self):
         ret = ""
-        for k in sorted(self.distribution.keys()):
+        for k in sorted(self.bins):
             ret += str(round(k,2)) + ':\t'
             if self.type == "histogram":
                 ret +=  str(round(self.distribution[k]  / self.count,3))
+            elif self.type == "KDE":
+                ret +=  str(round(self.density(k),3))
             else:
                 ret += str(round(self.distribution[k], 6))
             ret += '\n'

pyFTS/tests/general.py

@@ -15,9 +15,18 @@ from pyFTS.data import TAIEX
 
 dataset = TAIEX.get_data()
 
-from pyFTS.benchmarks import benchmarks as bchmk, Util as bUtil, Measures
+from pyFTS.benchmarks import benchmarks as bchmk, Util as bUtil, Measures, knn, quantreg, arima
+
+from pyFTS.models import pwfts, song, ifts
+
+model = arima.ARIMA("", order=(1,0,0))
+model.fit(dataset[:800])
+tmp = model.predict(dataset[800:1000], type='distribution')
+for tmp2 in tmp:
+    print(tmp2)
+
+
 
-from pyFTS.models import pwfts, song
 '''
 from pyFTS.partitioners import Grid, Util as pUtil
 partitioner = Grid.GridPartitioner(data=dataset[:800], npart=10, transformation=tdiff)
@@ -31,28 +40,39 @@ print(Measures.get_distribution_statistics(dataset[800:1000], model, steps_ahead
 #    print(tmp2)
 '''
 
-#'''
+'''
 
 from pyFTS.benchmarks import arima, naive, quantreg
 
-bchmk.sliding_window_benchmarks(dataset[:1000], 1000, train=0.8, inc=0.2,
+bchmk.sliding_window_benchmarks(dataset, 1000, train=0.8, inc=0.2,
-                                #methods=[song.ConventionalFTS], #[pwfts.ProbabilisticWeightedFTS],
+                                methods=[ifts.IntervalFTS], #[pwfts.ProbabilisticWeightedFTS],
                                 benchmark_models=True,
-                                benchmark_methods=[naive.Naive, arima.ARIMA,arima.ARIMA], #arima.ARIMA,arima.ARIMA],
+                                benchmark_methods=[arima.ARIMA for k in range(8)]
-                                #benchmark_methods=[arima.ARIMA],
+                                    + [quantreg.QuantileRegression for k in range(4)],
-                                benchmark_methods_parameters=[1,(1,0,0),(1,0,1)], #(2,0,1),(2,0,2)],
+                                benchmark_methods_parameters=[
-                                #benchmark_methods_parameters=[(1,0,0)],
+                                    {'order': (1, 0, 0), 'alpha': .05},
+                                    {'order': (1, 0, 0), 'alpha': .25},
+                                    {'order': (1, 0, 1), 'alpha': .05},
+                                    {'order': (1, 0, 1), 'alpha': .25},
+                                    {'order': (2, 0, 1), 'alpha': .05},
+                                    {'order': (2, 0, 1), 'alpha': .25},
+                                    {'order': (2, 0, 2), 'alpha': .05},
+                                    {'order': (2, 0, 2), 'alpha': .25},
+                                    {'order': 1, 'alpha': .05},
+                                    {'order': 1, 'alpha': .25},
+                                    {'order': 2, 'alpha': .05},
+                                    {'order': 2, 'alpha': .25},
+                                ],
                                 transformations=[None, tdiff],
-                                orders=[1, 2, 3],
+                                orders=[1], #2, 3],
-                                partitions=[35], #np.arange(10, 100, 5),
+                                partitions=[3], #np.arange(3, 25, 2),
-                                progress=True, type='point',
+                                progress=False, type='interval',
                                 #steps_ahead=[1,4,7,10], #steps_ahead=[1]
-                                #distributed=True, nodes=['192.168.0.110', '192.168.0.105','192.168.0.106'],
+                                distributed=True, nodes=['192.168.0.110', '192.168.0.107','192.168.0.106'],
-                                file="benchmarks.tmp", dataset="TAIEX", tag="comparisons")
+                                file="benchmarks.db", dataset="TAIEX", tag="comparisons")
-                                #save=True, file="tmp.db")
 
 
-#'''
+'''
 '''
 dat = pd.read_csv('pwfts_taiex_partitioning.csv', sep=';')
 print(bUtil.analytic_tabular_dataframe(dat))
@@ -111,4 +131,4 @@ tmp[20].plot(ax[2][2], title='t=200')
 f, ax = plt.subplots(1, 1, figsize=[20,15])
 bchmk.plot_distribution(ax, 'blue', tmp, f, 0, reference_data=dataset[train_split:train_split+200])
 
-'''
+'''