Include information about model performance in easier to access format
Nobody has claimed this yet.
Assessment
- Difficulty
- 5/5
- Estimated time
- Over a week
- Newbie friendliness
- 35/100
- Issue type
- Feature
- Clarity
- Needs clarification
- Activity status
- Stale
- Tech stack
- python
- Domain
- data, machine-learning
Research direction
Start by examining the existing data components and the general to_df() output to understand which forecast, observed, detrended, holiday, country, and date values are already available. Compare that structure with the proposed performance_df() option and the listed analysis questions. Done means a consistent DataFrame format exposes the requested absolute and relative comparisons, with its intended scope distinguished from Prophet's performance_metrics.
Written by the indexing model from the issue text.
Description
I think this is something that can be calculated from existing data components, but might not currently be packaged up in one consistent format.
Here were some of the analysis questions I encountered that I could not immediately figure out how to address with the current data structure:
- Was the effect of X holiday larger or smaller this year than prior year(s)? By how much? (In absolute count and in relative percent)
- When comparing trends (i.e. across multiple dates), is the gap between forecast and observed growing, shrinking, or staying level?
- E.g., if a model underestimated holiday impact, do the trends re-converge in the following days or does that gap persist as a step-change?
- Within each country/population, how far off was the forecast value from the observed value on a given date? (In absolute count and relative percent)
- What exactly was the impact of holiday(s) on the forecast value for a given date? (In absolute count and relative percent)
- This may just be the difference between the
detrendedforecast and the final forecast.
- This may just be the difference between the
- What was the estimated impact of holiday(s) on a historical observed value for a given date? (In absolute count and relative percent)
- This may just be the difference between the
detrendedamount and the observed amount.
- This may just be the difference between the
Ideally, this could be included in a simple DataFrame -- either as additional columns in the same general DataFrame output from the general to_df() function, or maybe as a separate dedicated function for performance_df().
(Also acknowledging that this might be different than some other object that stores overall measures of model performance -- things that might come out of Prophet's performance_metrics).
- Dominant language
- Python
- Stars
- 3
- Forks
- 3
- PR merge metrics
- No merged PRs in 30d
Getting set up
This project ships no dev container, Dockerfile or contributing guide, so setting up is up to you: start from its README, and see our first-contribution guide for the general steps.
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
More from mozilla/mozaic-forecasting
-
Difficulty 5/5 Over a week Newbie friendliness 30/100
-
Difficulty 4/5 3-5 days Newbie friendliness 35/100
-
Difficulty 5/5 Over a week Newbie friendliness 30/100
-
Difficulty 2/5 1-3 hours Newbie friendliness 45/100
All issues in mozilla/mozaic-forecasting
Similar issues
-
Difficulty 1/5 Under an hour Newbie friendliness 60/100
521xueweihan/HelloGitHub#3924 ·
-
Difficulty 2/5 1-3 hours Newbie friendliness 67/100
wilbowes/EchoMuse#869 · 1 comment ·
Maintainers usually reply within 1 day
-
Difficulty 1/5 Under an hour Newbie friendliness 85/100
-
namespace operations
Difficulty 1/5 Under an hour Newbie friendliness 72/100
EclipseFdn/open-vsx.org#14043 ·
Maintainers usually reply within 1 day
-
test: TestServeUntilStale races the server's close against the client's sendall (BrokenPipeError under load)Possibly taken @evoludigit claimed this today. Open
Difficulty 1/5 Under an hour Newbie friendliness 89/100
Maintainers usually reply within 1 day