From 88e788cdcae7178451337fa8a6611ae5b6895d40 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Petr=C3=B4nio=20C=C3=A2ndido?= <petronio.candido@gmail.com>
Date: Mon, 24 Dec 2018 09:33:36 -0200
Subject: [PATCH] Tsaur model

---
 pyFTS/common/flrg.py   |   2 +-
 pyFTS/models/tsaur.py  | 117 +++++++++++++++++++++++++++++++++++++++++
 pyFTS/tests/general.py |  24 ++++++---
 3 files changed, 134 insertions(+), 9 deletions(-)
 create mode 100644 pyFTS/models/tsaur.py

diff --git a/pyFTS/common/flrg.py b/pyFTS/common/flrg.py
index 3c9786e..11dcfc3 100644
--- a/pyFTS/common/flrg.py
+++ b/pyFTS/common/flrg.py
@@ -73,7 +73,7 @@ class FLRG(object):
         if isinstance(self.RHS, (list, set)):
             return np.array([sets[s].centroid for s in self.RHS])
         elif isinstance(self.RHS, dict):
-            return np.array([sets[self.RHS[s]].centroid for s in self.RHS.keys()])
+            return np.array([sets[s].centroid for s in self.RHS.keys()])
 
     def get_lower(self, sets):
         """
diff --git a/pyFTS/models/tsaur.py b/pyFTS/models/tsaur.py
new file mode 100644
index 0000000..539d1cd
--- /dev/null
+++ b/pyFTS/models/tsaur.py
@@ -0,0 +1,117 @@
+"""
+First order markov chain weighted FTS
+
+Tsaur, Ruey-Chyn. A FUZZY TIME SERIES-MARKOV CHAIN MODEL WITH AN APPLICATION TO FORECAST THE EXCHANGE RATE
+BETWEEN THE TAIWAN AND US DOLLAR. International Journal of Innovative Computing, Information and Control
+vol 8, no 7(B), p. 4931–4942, 2012.
+"""
+
+
+import numpy as np
+from pyFTS.common import FuzzySet,FLR,fts, flrg
+
+
+class MarkovWeightedFLRG(flrg.FLRG):
+    """First Order Markov Chain Weighted Fuzzy Logical Relationship Group"""
+    def __init__(self, LHS, **kwargs):
+        super(MarkovWeightedFLRG, self).__init__(1, **kwargs)
+        self.LHS = LHS
+        self.RHS = {}
+        self.count = 0.0
+        self.w = None
+
+    def append_rhs(self, c, **kwargs):
+        count = kwargs.get('count', 1.0)
+        if c not in self.RHS:
+            self.RHS[c] = count
+        else:
+            self.RHS[c] += count
+        self.count += count
+
+    def weights(self):
+        if self.w is None:
+            self.w = np.array([ v/self.count for k,v in self.RHS.items()])
+        return self.w
+
+    def get_midpoint(self, sets):
+        mp = np.array([sets[c].centroid for c in self.RHS.keys()])
+        return mp.dot(self.weights())
+
+    def __str__(self):
+        tmp = self.LHS + " -> "
+        tmp2 = ""
+        for c in sorted(self.RHS.keys()):
+            if len(tmp2) > 0:
+                tmp2 = tmp2 + ", "
+            tmp2 = tmp2 + c + "(" + str(self.RHS[c]/self.count) + ")"
+        return tmp + tmp2
+
+    def __len__(self):
+        return len(self.RHS)
+
+
+class MarkovWeightedFTS(fts.FTS):
+    """First Order Markov Chain Weighted Fuzzy Time Series"""
+    def __init__(self, **kwargs):
+        super(MarkovWeightedFTS, self).__init__(order=1, name="MWFTS", **kwargs)
+        self.name = "Markov Weighted FTS"
+        self.detail = "Tsaur"
+        self.is_high_order = False
+        self.order = 1
+
+    def generate_flrg(self, flrs):
+        for flr in flrs:
+            if flr.LHS in self.flrgs:
+                self.flrgs[flr.LHS].append_rhs(flr.RHS)
+            else:
+                self.flrgs[flr.LHS] = MarkovWeightedFLRG(flr.LHS)
+                self.flrgs[flr.LHS].append_rhs(flr.RHS)
+
+    def train(self, data, **kwargs):
+        tmpdata = FuzzySet.fuzzyfy(data, partitioner=self.partitioner, method='maximum', mode='sets')
+        flrs = FLR.generate_recurrent_flrs(tmpdata)
+        self.generate_flrg(flrs)
+
+    def forecast(self, ndata, **kwargs):
+
+        explain = kwargs.get('explain', False)
+
+        if self.partitioner is not None:
+            ordered_sets = self.partitioner.ordered_sets
+        else:
+            ordered_sets = FuzzySet.set_ordered(self.sets)
+
+        data = np.array(ndata)
+
+        l = len(ndata)
+
+        ret = []
+
+        for k in np.arange(0, l):
+
+            actual = FuzzySet.get_maximum_membership_fuzzyset(ndata[k], self.sets, ordered_sets)
+
+            if explain:
+                print("Fuzzyfication:\n\n {} -> {} \n".format(ndata[k], actual.name))
+
+            if actual.name not in self.flrgs:
+                ret.append(actual.centroid)
+
+                if explain:
+                    print("Rules:\n\n {} -> {} (Naïve)\t Midpoint: {}  \n\n".format(actual.name, actual.name,actual.centroid))
+
+            else:
+                flrg = self.flrgs[actual.name]
+                mp = flrg.get_midpoints(self.sets)
+
+                final = mp.dot(flrg.weights())
+
+                ret.append(final)
+
+                if explain:
+                    print("Rules:\n\n {} \n\n ".format(str(flrg)))
+                    print("Midpoints: \n\n {}\n\n".format(mp))
+
+                    print("Deffuzyfied value: {} \n".format(final))
+
+        return ret
diff --git a/pyFTS/tests/general.py b/pyFTS/tests/general.py
index 8ef7d8b..9f28d5f 100644
--- a/pyFTS/tests/general.py
+++ b/pyFTS/tests/general.py
@@ -19,22 +19,30 @@ tdiff = Transformations.Differential(1)
 
 from pyFTS.data import TAIEX, SP500, NASDAQ, Malaysia, Enrollments
 
-from pyFTS.data import mackey_glass
-y = mackey_glass.get_data()
+#from pyFTS.data import mackey_glass
+#y = mackey_glass.get_data()
 
 from pyFTS.partitioners import Grid
-from pyFTS.models import pwfts
+from pyFTS.models import pwfts, tsaur
 
-partitioner = Grid.GridPartitioner(data=y, npart=35)
+train = TAIEX.get_data()[:1000]
+test = TAIEX.get_data()[1000:1200]
 
-model = pwfts.ProbabilisticWeightedFTS(partitioner=partitioner, order=2, lags=[3,4])
-model.fit(y[:800])
+partitioner = Grid.GridPartitioner(data=train, npart=35)
+
+#model = pwfts.ProbabilisticWeightedFTS(partitioner=partitioner) #, order=2, lags=[3,4])
+model = tsaur.MarkovWeightedFTS(partitioner=partitioner)
+model.fit(train)
 
 from pyFTS.benchmarks import benchmarks as bchmk
 
-distributions = model.predict(y[800:820])
+print(model)
 
-print(distributions)
+print(model.forecast(test))
+
+#distributions = model.predict(y[800:820])
+
+#print(distributions)
 
 
 '''