Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Chronos: Add support for id_sensitive=True to Forecaster.from_tsdataset #5551

Merged
merged 5 commits into from
Sep 10, 2022
+16 −3
Merged

Chronos: Add support for id_sensitive=True to Forecaster.from_tsdataset #5551

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
bc812fa
supports id_sensitive=True
liangs6212 Aug 26, 2022
d6a3c4d
new params is_predict
liangs6212 Aug 30, 2022
04f689e
rollback is_predict
liangs6212 Aug 30, 2022
ee978e0
fix known issues
liangs6212 Sep 3, 2022
ad7021d
fix code style
liangs6212 Sep 3, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 6 additions & 1 deletion python/chronos/src/bigdl/chronos/forecaster/base_forecaster.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1096,7 +1096,7 @@ def from_tsdataset(cls, tsdataset, past_seq_len=None, future_seq_len=None, **kwa
"""
Build a Forecaster Model.

:param tsdataset: A bigdl.chronos.data.tsdataset.TSDataset instance.
:param tsdataset: Train tsdataset, a bigdl.chronos.data.tsdataset.TSDataset instance.
:param past_seq_len: int or "auto", Specify the history time steps (i.e. lookback).
Do not specify the 'past_seq_len' if your tsdataset has called
the 'TSDataset.roll' method or 'TSDataset.to_torch_data_loader'.
Expand Down Expand Up @@ -1151,6 +1151,11 @@ def check_time_steps(tsdataset, past_seq_len, future_seq_len):
fixMsg="Do not specify past_seq_len and future seq_len "
"or call tsdataset.roll method again and specify time step")

if tsdataset.id_sensitive:
_id_list_len = len(tsdataset.id_col)
input_feature_num *= _id_list_len
output_feature_num *= _id_list_len

return cls(past_seq_len=past_seq_len,
future_seq_len=future_seq_len,
input_feature_num=input_feature_num,
Expand Down
7 changes: 6 additions & 1 deletion python/chronos/src/bigdl/chronos/forecaster/lstm_forecaster.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -152,7 +152,7 @@ def from_tsdataset(cls, tsdataset, past_seq_len=None, **kwargs):
'''
Build a LSTM Forecaster Model.

:param tsdataset: A bigdl.chronos.data.tsdataset.TSDataset instance.
:param tsdataset: Train tsdataset, a bigdl.chronos.data.tsdataset.TSDataset instance.
:param past_seq_len: Specify the history time steps (i.e. lookback).
Do not specify the 'past_seq_len' if your tsdataset has called
the 'TSDataset.roll' method or 'TSDataset.to_torch_data_loader'.
Expand Down Expand Up @@ -198,6 +198,11 @@ def check_time_steps(tsdataset, past_seq_len):
fixMsg="Do not specify past_seq_len "
"or call tsdataset.roll method again and specify time step.")

if tsdataset.id_sensitive:
_id_list_len = len(tsdataset._id_list)
input_feature_num *= _id_list_len
output_feature_num *= _id_list_len

return cls(past_seq_len=past_seq_len,
input_feature_num=input_feature_num,
output_feature_num=output_feature_num,
Expand Down
5 changes: 4 additions & 1 deletion python/chronos/src/bigdl/chronos/forecaster/nbeats_forecaster.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -174,7 +174,7 @@ def from_tsdataset(cls, tsdataset, past_seq_len=None, future_seq_len=None, **kwa
"""
Build a NBeats Forecaster Model.

:param tsdataset: A bigdl.chronos.data.tsdataset.TSDataset instance.
:param tsdataset: Train tsdataset, a bigdl.chronos.data.tsdataset.TSDataset instance.
:param past_seq_len: Specify the history time steps (i.e. lookback).
Do not specify the 'past_seq_len' if your tsdataset has called
the 'TSDataset.roll' method or 'TSDataset.to_torch_data_loader'.
Expand Down Expand Up @@ -225,6 +225,9 @@ def check_time_steps(tsdataset, past_seq_len, future_seq_len):
fixMsg="Do not specify past_seq_len and future seq_len "
"or call tsdataset.roll method again and specify time step")

invalidInputError(not all([tsdataset.id_sensitive, len(tsdataset._id_list) > 1]),
"NBeats only supports univariate forecasting.")

return cls(past_seq_len=past_seq_len,
future_seq_len=future_seq_len,
**kwargs)