Bug 206 #208
Bug 206 #208
Conversation
- Added checks for missing performance data in generate_report.py. - Updated ConfigurationOutput class to track performance data availability and reasons. - Improved error handling in PerformanceDataFrame for missing measurements. - Introduced parsing for unquoted IRACE categorical values in PCSConverter. - Added tests for unquoted categorical values and enhanced test coverage.
…elector metadata handling
|
Update the changeling with a new line describing the bug you've fixed |
thijssnelleman
left a comment
thijssnelleman
left a comment
There was a problem hiding this comment.
Overall looks good, we should discuss a few changes in the office
src/sparkle/CLI/generate_report.py
Outdated
| ) | ||
| report.append(table_status_values) | ||
|
|
||
| if scenario_output.performance_data_missing: |
There was a problem hiding this comment.
This should never occur, in what situation does this happen?
There was a problem hiding this comment.
Also, instead of creating a new property on the scenario, you can just do scenario_output.performance_data.exists() (just calling .exists on the path will give you this information directly?)
src/sparkle/tools/parameters.py
Outdated
| ) | ||
|
|
||
| @staticmethod | ||
| def parse_irace_categorical_values(raw_values: str) -> list[str]: |
There was a problem hiding this comment.
This makes sense, but shouldn't this be defined within the IRACE parse function? Also can we use regex instead? I know this wasn't done a lot yet in this file but regex are overall faster, less error prone and easier to read
There was a problem hiding this comment.
You're right! I'll also switch to regex.
…issing logic, bring settings_object to base version
…_selection.sh and streamline categorical value parsing in PCSConverter
…o configuration-space parsing and performance data cleaning
Adjusting latex.py to eliminate SciPy warnings/errors and improving data handling
Adding parser to irace for parsing categorical values in tools/parameters.py
Adding tests for configspace.py
Handling only NaN values containing performance dataframe case