Bugfix in Transformations.ROI

pyFTS/common/Transformations.py

@@ -250,9 +250,7 @@ class ROI(Transformation):
         return modified
 
     def inverse(self, data, param=None, **kwargs):
-        modified = [param[0]]
-        for i in np.arange(1, len(data)):
-            modified.append((modified[i - 1] * data[i]) + modified[i - 1])
+        modified = [(param[i - 1] * data[i]) + param[i - 1] for i in np.arange(1, len(data))]
         return modified
 
 

pyFTS/tests/transformations.py

@@ -53,7 +53,7 @@ detrend = trend.apply(data)
 plt.plot(trend.inverse(detrend, data, date_offset=pd.DateOffset(months=1)))
 '''
 
-
+'''
 data = Enrollments.get_dataframe()
 
 trend = Transformations.LinearTrend(data_field='Enrollments', index_field='Year')
@@ -67,8 +67,64 @@ plt.plot(trend.trend(data)) #)
 detrend = trend.apply(data)
 
 plt.plot(trend.inverse(detrend, data))
+'''
+
+dataset = pd.read_csv('https://query.data.world/s/nxst4hzhjrqld4bxhbpn6twmjbwqk7')
+dataset['data'] = pd.to_datetime([str(y)+'-'+str(m) for y,m in zip(dataset['Ano'].values, dataset['Mes'].values)],
+                                  format='%Y-%m')
+roi = Transformations.ROI()
+
+
+train = dataset['Total'].values[:30]
+test = dataset['Total'].values[30:]
+
+fs = Grid.GridPartitioner(data=train, npart=5, transformation=roi)
+
+from pyFTS.models import hofts, pwfts
+
+model = pwfts.ProbabilisticWeightedFTS(partitioner=fs, order=2)
+#model = hofts.WeightedHighOrderFTS(partitioner=fs, order=1)
+model.append_transformation(roi)
+
+
+model.fit(train)
+
+fig, ax = plt.subplots(nrows=1, ncols=1, figsize=[10,5])
+ax.plot(test)
+
+'''
+train = dataset.iloc[:30]
+test = dataset.iloc[30:]
+
+from pyFTS.models.multivariate import common, variable, mvfts, wmvfts
+from pyFTS.partitioners import Grid, Entropy
+from pyFTS.models.seasonal.common import DateTime
+from pyFTS.models.seasonal import partitioner as seasonal
+sp = {'seasonality': DateTime.month , 'names': ['Jan','Fev','Mar','Abr','Mai','Jun','Jul', 'Ago','Set','Out','Nov','Dez']}
+
+vmonth = variable.Variable("Month", data_label="data", partitioner=seasonal.TimeGridPartitioner, npart=12,
+                           data=train, partitioner_specific=sp)
+
+vtur = variable.Variable("Turistas", data_label="Total", alias='tur',
+                         partitioner=Grid.GridPartitioner, npart=10, transformation=roi,
+                         data=train)
+
+model = wmvfts.WeightedMVFTS(explanatory_variables=[vmonth, vtur], target_variable=vtur)
+model.fit(train)
+
+fig, ax = plt.subplots(nrows=1, ncols=1, figsize=[10,5])
+
+ax.plot(test['Total'].values)
+'''
+
+forecast = model.predict(test)
+
+for k in np.arange(model.order):
+  forecast.insert(0,None)
+
+ax.plot(forecast)
 
 
 plt.show()
 
-print(data)
+print(dataset)