sam/pyFTS
1
0
Fork 0
Code Issues 3 Pull Requests Actions Packages Projects Releases Wiki Activity

Bugfix: GranularMVFTS and ClusteredMVFTS were not considering data transformations

Browse Source
This commit is contained in:
Petrônio Cândido 2020-01-27 12:09:54 -03:00
parent 94499f0f21
commit 783a77ec0f
2 changed files with 18 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
pyFTS/models/multivariate/cmvfts.py
View File

@ -50,7 +50,9 @@ class ClusteredMVFTS(mvfts.MVFTS):
self.model = self.fts_method(partitioner=self.partitioner, **self.fts_params) self.model = self.fts_method(partitioner=self.partitioner, **self.fts_params)
ndata = self.check_data(data) ndata = self.apply_transformations(data)
ndata = self.check_data(ndata)
self.model.train(ndata, fuzzyfied=self.pre_fuzzyfy) self.model.train(ndata, fuzzyfied=self.pre_fuzzyfy)
@ -64,13 +66,20 @@ class ClusteredMVFTS(mvfts.MVFTS):
return ndata return ndata
def forecast(self, ndata, **kwargs): def forecast(self, data, **kwargs):
ndata = self.check_data(ndata) ndata1 = self.apply_transformations(data)
ndata = self.check_data(ndata1)
pre_fuzz = kwargs.get('pre_fuzzyfy', self.pre_fuzzyfy) pre_fuzz = kwargs.get('pre_fuzzyfy', self.pre_fuzzyfy)
return self.model.forecast(ndata, fuzzyfied=pre_fuzz, **kwargs) ret = self.model.forecast(ndata, fuzzyfied=pre_fuzz, **kwargs)
ret = self.target_variable.apply_inverse_transformations(ret,
params=data[self.target_variable.data_label].values)
return ret
def forecast_interval(self, data, **kwargs): def forecast_interval(self, data, **kwargs):

9
pyFTS/tests/transformations.py
View File

@ -74,6 +74,7 @@ dataset['data'] = pd.to_datetime([str(y)+'-'+str(m) for y,m in zip(dataset['Ano'
format='%Y-%m') format='%Y-%m')
roi = Transformations.ROI() roi = Transformations.ROI()
'''
train = dataset['Total'].values[:30] train = dataset['Total'].values[:30]
test = dataset['Total'].values[30:] test = dataset['Total'].values[30:]
@ -96,7 +97,7 @@ ax.plot(test)
train = dataset.iloc[:30] train = dataset.iloc[:30]
test = dataset.iloc[30:] test = dataset.iloc[30:]
from pyFTS.models.multivariate import common, variable, mvfts, wmvfts from pyFTS.models.multivariate import common, variable, mvfts, wmvfts, granular
from pyFTS.partitioners import Grid, Entropy from pyFTS.partitioners import Grid, Entropy
from pyFTS.models.seasonal.common import DateTime from pyFTS.models.seasonal.common import DateTime
from pyFTS.models.seasonal import partitioner as seasonal from pyFTS.models.seasonal import partitioner as seasonal
@ -106,16 +107,16 @@ vmonth = variable.Variable("Month", data_label="data", partitioner=seasonal.Time
data=train, partitioner_specific=sp) data=train, partitioner_specific=sp)
vtur = variable.Variable("Turistas", data_label="Total", alias='tur', vtur = variable.Variable("Turistas", data_label="Total", alias='tur',
partitioner=Grid.GridPartitioner, npart=10, transformation=roi, partitioner=Grid.GridPartitioner, npart=20, transformation=roi,
data=train) data=train)
model = wmvfts.WeightedMVFTS(explanatory_variables=[vmonth, vtur], target_variable=vtur) #model = wmvfts.WeightedMVFTS(explanatory_variables=[vmonth, vtur], target_variable=vtur)
model = granular.GranularWMVFTS(explanatory_variables=[vmonth, vtur], target_variable=vtur, order=2, knn=1)
model.fit(train) model.fit(train)
fig, ax = plt.subplots(nrows=1, ncols=1, figsize=[10,5]) fig, ax = plt.subplots(nrows=1, ncols=1, figsize=[10,5])
ax.plot(test['Total'].values) ax.plot(test['Total'].values)
'''
forecast = model.predict(test) forecast = model.predict(test)