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

Improvements for forecasting ahead in ClusteredMVFTS

Browse Source
This commit is contained in:
Petrônio Cândido 2019-06-21 15:40:20 -03:00
parent a6d9d164e4
commit de82af1b9d
3 changed files with 16 additions and 16 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
pyFTS/models/multivariate/cmvfts.py
View File

@ -109,10 +109,10 @@ class ClusteredMVFTS(mvfts.MVFTS):
ndata = self.apply_transformations(data) ndata = self.apply_transformations(data)
start = kwargs.get('start_at', self.order) start = kwargs.get('start_at', 0)
ret = [] ret = []
sample = ndata.iloc[start - self.max_lag:] sample = ndata.iloc[start: start + self.max_lag]
for k in np.arange(0, steps): for k in np.arange(0, steps):
tmp = self.forecast_distribution(sample.iloc[-self.max_lag:], **kwargs)[0] tmp = self.forecast_distribution(sample.iloc[-self.max_lag:], **kwargs)[0]

26
pyFTS/models/multivariate/mvfts.py
View File

@ -170,13 +170,12 @@ class MVFTS(fts.FTS):
ndata = self.apply_transformations(data) ndata = self.apply_transformations(data)
start = kwargs.get('start_at', self.max_lag) start = kwargs.get('start_at', 0)
ndata = ndata.loc[ndata.index[start-self.max_lag:start]] ndata = ndata.iloc[start: start + self.max_lag]
ret = [] ret = []
for k in np.arange(start, start+steps): for k in np.arange(0, steps):
ix = ndata.index[k-self.max_lag:k] sample = ndata.iloc[-self.max_lag:]
sample = ndata.loc[ix]
tmp = self.forecast(sample, **kwargs) tmp = self.forecast(sample, **kwargs)
if isinstance(tmp, (list, np.ndarray)): if isinstance(tmp, (list, np.ndarray)):
@ -192,12 +191,13 @@ class MVFTS(fts.FTS):
last_data_point = ndata.loc[ndata.index[-1]] last_data_point = ndata.loc[ndata.index[-1]]
new_data_point[data_label] = generators[data_label](last_data_point[data_label]) new_data_point[data_label] = generators[data_label](last_data_point[data_label])
elif isinstance(generators[data_label], fts.FTS): elif isinstance(generators[data_label], fts.FTS):
model = generators[data_label] gen_model = generators[data_label]
last_data_point = ndata.loc[[ndata.index[-model.order]]] last_data_point = sample.iloc[-gen_model.order:]
if not model.is_multivariate:
if not gen_model.is_multivariate:
last_data_point = last_data_point[data_label].values last_data_point = last_data_point[data_label].values
new_data_point[data_label] = model.forecast(last_data_point)[0] new_data_point[data_label] = gen_model.forecast(last_data_point)[0]
new_data_point[self.target_variable.data_label] = tmp new_data_point[self.target_variable.data_label] = tmp
@ -260,10 +260,10 @@ class MVFTS(fts.FTS):
ndata = self.apply_transformations(data) ndata = self.apply_transformations(data)
start = kwargs.get('start_at', self.order) start = kwargs.get('start_at', 0)
ret = [] ret = []
ix = ndata.index[start - self.max_lag:] ix = ndata.index[start: start + self.max_lag]
lo = ndata.loc[ix] #[ndata.loc[k] for k in ix] lo = ndata.loc[ix] #[ndata.loc[k] for k in ix]
up = ndata.loc[ix] #[ndata.loc[k] for k in ix] up = ndata.loc[ix] #[ndata.loc[k] for k in ix]
for k in np.arange(0, steps): for k in np.arange(0, steps):
@ -284,8 +284,8 @@ class MVFTS(fts.FTS):
new_data_point_up[data_label] = generators[data_label](last_data_point_up[data_label]) new_data_point_up[data_label] = generators[data_label](last_data_point_up[data_label])
elif isinstance(generators[data_label], fts.FTS): elif isinstance(generators[data_label], fts.FTS):
model = generators[data_label] model = generators[data_label]
last_data_point_lo = lo.loc[lo.index[-model.order]] last_data_point_lo = lo.loc[lo.index[-model.order:]]
last_data_point_up = up.loc[up.index[-model.order]] last_data_point_up = up.loc[up.index[-model.order:]]
if not model.is_multivariate: if not model.is_multivariate:
last_data_point_lo = last_data_point_lo[data_label].values last_data_point_lo = last_data_point_lo[data_label].values

2
pyFTS/tests/pwfts.py
View File

@ -48,7 +48,7 @@ model = granular.GranularWMVFTS(explanatory_variables=[vhour, vtemp, vload], tar
model.fit(train_mv) model.fit(train_mv)
temp_generator = pwfts.ProbabilisticWeightedFTS(partitioner=vtemp.partitioner, order=2) temp_generator = pwfts.ProbabilisticWeightedFTS(partitioner=vtemp.partitioner, order=1)
temp_generator.fit(train_mv['temperature'].values) temp_generator.fit(train_mv['temperature'].values)
#print(model) #print(model)