Form new beliefs given some model #50
Conversation
Ridgeline
* added generate function * Update generate.py * Create ridgeline plot * Add example plots ridgeline * add ridgeline examples * add titles * added plotting function and cleaned up code * added plotting and cleaned up code * code cleanup and commenting * fixed typo's * update + error messages * updated error messages * Copied ridgeline README documentation * Added ridgeline README files * Added generate documentation * fixed typo * fix labels
Distinguish how mean and median averages are calculated. Allow additional distribution parameters to be set.
|
Test setup is failing for Python 3.7, with |
# Conflicts: # setup.py
…opy statement now that we explicitly test for input being unaffected
timely_beliefs/beliefs/classes.py
Outdated
| freq=df.event_resolution, | ||
| closed="left", | ||
| ) | ||
| forecaster.fit(df["event_value"]) |
There was a problem hiding this comment.
At this time, our BeliefDataFrames do not support regressors to be used here, correct? For example, forecasting wind speed based on time of day and temperature - how would we do that? (maybe another ticket)
There was a problem hiding this comment.
Good idea to separate this into another ticket. It can be done by passing an extra X to the fit function, for example, like this:
X=df.loc[:, df.columns != "event_value"]
Here, I'm assuming the external regressors are given as additional columns in the BeliefsDataFrame.
Testing it with sktime's naive forecaster will just lead to the forecaster ignoring X though, so that doesn't feel like a wholesome test.
There was a problem hiding this comment.
Did we make an issue out of this yet? I can't find one.
There was a problem hiding this comment.
Ah, I see you already add the X in this PR. I believe a little comment would help.
And is there a case where columns in X would be bad to include?
We could also add an optional parameter to form_beliefs, e.g. "X_column_names".
There was a problem hiding this comment.
I created #116 just now.
…contain any records prior to belief_time
# Conflicts: # timely_beliefs/beliefs/utils.py
This introduces a new BeliefsDataFrame method
form_beliefsthat creates new beliefs given amodel, abelief_timeand anevent_time_window.Much credit goes to the AI Master students from the University of Amsterdam that also helped us implement ridgeline plots for probabilistic forecasts.