-
Notifications
You must be signed in to change notification settings - Fork 234
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.
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
feat: Add CGAN for timeseries #108
base: dev
Are you sure you want to change the base?
Conversation
@@ -1,45 +1,46 @@ | |||
# Inverts all preprocessing pipelines provided in the preprocessing examples |
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.
Changes in this script consist in extending the inverse support to the MinMaxScaler
@@ -2,17 +2,30 @@ | |||
Get the stock data from Yahoo finance data |
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.
Allowing subsetting of columns
@@ -4,7 +4,7 @@ | |||
import numpy as np |
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.
Enabling inverse_transform of the timeseries scaler (processor needs to be returned here)
src/ydata_synthetic/postprocessing/timeseries/inverse_preprocesser.py
Outdated
Show resolved
Hide resolved
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.
Revise sample method.
remove changes on gitignore removed unused n_feats argument
apply revisions
Auto regressive timeseries sampling method revert TS data processor integration
efa508a
to
a3c9b34
Compare
data_.append(records) | ||
data_ = hstack(data_)[:, :seq_len] | ||
data.append(data_) | ||
return array(vstack(data)) |
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.
Regarding to a revision of this sample method I think:
- We can force the number of samples to be the same as requested, i.e. truncate the last batch if needed
- Condition converted to tensor could be taken outside the for loop
- It does not seem to be truly auto-regressive since only the condition part of the produced records is being used as condition, we can add this capacity
Do you agree/have anything to add @fabclmnt ?
Closes #93