From ac71ee34c63749577e370b60e14c5d098b3595a5 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Petr=C3=B4nio=20C=C3=A2ndido?= <petronio.candido@gmail.com>
Date: Thu, 30 May 2019 07:59:12 -0300
Subject: [PATCH] ensemble.SimpleEnsembleFTS

---
 pyFTS/models/ensemble/ensemble.py | 28 +++++++++++++++++
 pyFTS/tests/ensemble.py           | 52 +++++++++++++------------------
 2 files changed, 49 insertions(+), 31 deletions(-)

diff --git a/pyFTS/models/ensemble/ensemble.py b/pyFTS/models/ensemble/ensemble.py
index c4d5d62..c429ae3 100644
--- a/pyFTS/models/ensemble/ensemble.py
+++ b/pyFTS/models/ensemble/ensemble.py
@@ -12,6 +12,7 @@ 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
+from pyFTS.partitioners import Grid
 import scipy.stats as st
 from itertools import product
 
@@ -288,6 +289,33 @@ class EnsembleFTS(fts.FTS):
         return ret
 
 
+class SimpleEnsembleFTS(EnsembleFTS):
+    '''
+    An homogeneous FTS method ensemble with variations on partitionings and orders.
+    '''
+    def __init__(self, **kwargs):
+        super(SimpleEnsembleFTS, self).__init__(**kwargs)
+        self.method = kwargs.get('fts_method', hofts.WeightedHighOrderFTS)
+        """FTS method class that will be used on internal models"""
+        self.partitioner_method = kwargs.get('partitioner_method', Grid.GridPartitioner)
+        """UoD partitioner class that will be used on internal methods"""
+        self.partitions = kwargs.get('partitions', np.arange(15,35,10))
+        """Possible variations of number of partitions on internal models"""
+        self.orders = kwargs.get('orders', [1,2,3])
+        """Possible variations of order on internal models"""
+
+    def train(self, data, **kwargs):
+        for k in self.partitions:
+            fs = self.partitioner_method(data=data, npart=k)
+
+            for order in self.orders:
+                tmp = self.method(partitioner=fs, order=order)
+
+                tmp.fit(data)
+
+                self.append_model(tmp)
+
+
 class AllMethodEnsembleFTS(EnsembleFTS):
     """
     Creates an EnsembleFTS with all point forecast methods, sharing the same partitioner
diff --git a/pyFTS/tests/ensemble.py b/pyFTS/tests/ensemble.py
index f55de00..4ccc2cd 100644
--- a/pyFTS/tests/ensemble.py
+++ b/pyFTS/tests/ensemble.py
@@ -19,48 +19,38 @@ from pyFTS.models import hofts
 from pyFTS.data import TAIEX
 
 data = TAIEX.get_data()
-train = data[:800]
-test = data[800:1000]
-'''
-model = ensemble.EnsembleFTS()
 
-for k in [15, 25, 35]:
-    for order in [1, 2, 3]:
-        fs = Grid.GridPartitioner(data=train, npart=k)
-        tmp = hofts.WeightedHighOrderFTS(partitioner=fs, order=order)
 
-        tmp.fit(train)
 
-        model.append_model(tmp)
-'''
-#fig, ax = plt.subplots(nrows=1, ncols=1, figsize=[15, 5])
-
-#ax.plot(data[:28], label='Original', color='black')
+model1 = ensemble.SimpleEnsembleFTS()
 
 from pyFTS.benchmarks import arima, quantreg, BSTS
 
-#model = arima.ARIMA(order=(2,0,0))
-#model = quantreg.QuantileRegression(order=1, dist=True)
-model = BSTS.ARIMA(order=(2,0,0))
-model.fit(train)
+methods = [ensemble.SimpleEnsembleFTS, arima.ARIMA, quantreg.QuantileRegression, BSTS.ARIMA]
+parameters = [{},{'order': (2,0,0)}, {'order': 1, 'dist': True}, {'order': (2,0,0)}]
 
-horizon = 5
-
-intervals = model.predict(test[:10], type='interval', alpha=.25, steps_ahead=horizon)
-
-distributions = model.predict(test[:10], type='distribution', smooth='histogram', steps_ahead=horizon, num_bins=100)
-
-#Util.plot_distribution2(forecasts, data[:28], start_at=20, order=3, ax=ax, cmap="Blues")
-
-print('end')
 
 from pyFTS.benchmarks import Measures
 
-start = model.order+1
-end = start + horizon
+horizon = 5
+
+for ct, train, test, in Util.sliding_window(data,1000,0.8,.5):
+    print('data window {}'.format(ct))
+    for ct, method in enumerate(methods):
+        model = method(**parameters[ct])
+        model.fit(train)
+        start = model.order + 1
+        end = start + horizon
+        intervals = model.predict(test[:10], type='interval', alpha=.25, steps_ahead=horizon)
+        distributions = model.predict(test[:10], type='distribution', smooth='histogram', steps_ahead=horizon, num_bins=100)
+        print(model.name, Measures.get_interval_ahead_statistics(test[start:end], intervals))
+        print(model.name, Measures.get_distribution_ahead_statistics(test[start:end], distributions))
+
+print('end')
+
+
+
 
-print(Measures.get_interval_ahead_statistics(test[start:end], intervals))
-print(Measures.get_distribution_ahead_statistics(test[start:end], distributions))