Use uniform distribution in timeseries demo
#8020
Open
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
jrbourbeau
changed the title
timeseries demotimeseries demo
|
Consider me -0 on this FWIW I like having a poisson distribution. I think that it makes things feel a bit more natural, which is nice to see in random data. People can use However, if it's fast to produce "natural looking" data in some other faster way then I'd be for that. |
|
some related to extending the timeseries demo, @isVoid recently added a feature to generate nulls as part of the distribution: |
|
It is surprising to me how often I use the timeseries demo. I have never personally been bothered by the time it takes to generate the data, but I am all for making it even more useful. In case it's useful to others, I recently started adding an annual temperature column: annual_cycle = np.sin(2 * np.pi * (ddf.index.dayofyear.values / 365.25 - 0.28)).compute_chunk_sizes()
temperature_values = 10 15 * annual_cycle 3 * da.random.normal(size=annual_cycle.size)
ddf["temperature"] = temperature_values |
github-actions
bot
added
the
needs attention
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
I was putting together a demo using our
timeseriesutility and noticed generating random integers can take a significant amount of time -- especially when generating a DataFrame with many columns. After digging in a bit deeper, this is because we're generating integers from a Poisson distributiondask/dask/dataframe/io/demo.py
Lines 16 to 17 in da24582
which can be an order of magnitude slower than from, for example, a uniform distribution:
This PR proposes we switch to using a uniform distribution for generating random integers.
cc @rjzamora who has recently worked with our
timeseriesutility