-
Notifications
You must be signed in to change notification settings - Fork 80
Add AutoRegressivePipeline
#209
Conversation
message="You probably set wrong freq.", | ||
action="ignore", | ||
) | ||
cur_ts_forecast = cur_ts.make_future(cur_step) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Why can't we use self.ts to generate future?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It don't have all the values to make features for the future. And if we change it, it can make it difficult to make another forecast.
# Conflicts: # CHANGELOG.md
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Could you check behaviour of segmentencoder and inplace={False, True} Scalers too.
to_forecast = self.horizon | ||
while to_forecast > 0: | ||
cur_step = min(self.step, to_forecast) | ||
cur_ts = TSDataset(prediction_df, freq=self.ts.freq) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Let's use full name current_step
and current_ts
""" | ||
prediction_df = self.ts.to_pandas() | ||
to_forecast = self.horizon | ||
while to_forecast > 0: |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Let's make it with for loop
and all logic in one place:
for idx_start in range(0, total_lenght, batch_size):
len_slice = min(batch_size, total_lenght - idx_start)
|
||
Returns | ||
------- | ||
TSDataset |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Something strange.
There is above
Returns
-------
Pipeline:
Fitted Pipeline instance
) | ||
cur_ts_forecast = cur_ts.make_future(cur_step) | ||
cur_ts_future = self.model.forecast(cur_ts_forecast) | ||
prediction_df = pd.concat([prediction_df, cur_ts_future.to_pandas()[prediction_df.columns]]) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Can we just fill nans without concat every time?
# Conflicts: # CHANGELOG.md
Codecov Report
@@ Coverage Diff @@
## master #209 +/- ##
==========================================
+ Coverage 88.68% 88.98% +0.30%
==========================================
Files 77 78 +1
Lines 3631 3676 +45
==========================================
+ Hits 3220 3271 +51
+ Misses 411 405 -6
Continue to review full report at Codecov.
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
👍
IMPORTANT: Please do not create a Pull Request without creating an issue first.
Before submitting (must do checklist)
Type of Change
Proposed Changes
Add
AutoRegressivePipeline
.Related Issue
#173.
Closing issues
Closes #173.