Kuiper cleanup 10/n - Rename DEFAULT_MULTI_KUIPER constants to DEFAULT_MCE #169
Conversation
This file contains hidden or 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
flinder
requested review from
Lorenzo-Perini,
TaXxER and
danielha1
as code owners
meta-cla
bot
added
the
CLA Signed
flinder
added a commit
to flinder/MCGrad-1
that referenced
this pull request
Jan 23, 2026
…T_MCE (facebookincubator#169) Summary: Renames module-level constants, some variables and docstrings from legacy Kuiper naming to MCE (Multicalibration Error) naming for consistency with the ECCE API refactoring: - `DEFAULT_MULTI_KUIPER_MAX_VALUES_PER_SEGMENT_FEATURE` → `DEFAULT_MCE_MAX_VALUES_PER_SEGMENT_FEATURE` - `DEFAULT_MULTI_KUIPER_MIN_DEPTH` → `DEFAULT_MCE_MIN_DEPTH` - `DEFAULT_MULTI_KUIPER_MAX_DEPTH` → `DEFAULT_MCE_MAX_DEPTH` - `DEFAULT_MULTI_KUIPER_MIN_SAMPLES_PER_SEGMENT` → `DEFAULT_MCE_MIN_SAMPLES_PER_SEGMENT` - `DEFAULT_MULTI_KUIPER_GLOBAL_NORMALIZATION` → `DEFAULT_MCE_GLOBAL_NORMALIZATION` - `DEFAULT_MULTI_KUIPER_N_SEGMENTS` → `DEFAULT_MCE_N_SEGMENTS` Also updates fluent2 consumers to use the new constant names: - `fblearner/flow/projects/fluent2/definition/metrics/multicalibration_error.py` - `fblearner/flow/projects/fluent2/definition/evaluation.py` **Test Plan:** ``` buck2 test multicalibration/tests:test_metrics buck2 test fblearner/flow/projects/fluent2/workflows/evaluation/metrics/tests:multicalibration_error_test ``` Differential Revision: D91306073
flinder
added a commit
to flinder/MCGrad-1
that referenced
this pull request
Jan 23, 2026
…T_MCE (facebookincubator#169) Summary: Renames module-level constants, some variables and docstrings from legacy Kuiper naming to MCE (Multicalibration Error) naming for consistency with the ECCE API refactoring: - `DEFAULT_MULTI_KUIPER_MAX_VALUES_PER_SEGMENT_FEATURE` → `DEFAULT_MCE_MAX_VALUES_PER_SEGMENT_FEATURE` - `DEFAULT_MULTI_KUIPER_MIN_DEPTH` → `DEFAULT_MCE_MIN_DEPTH` - `DEFAULT_MULTI_KUIPER_MAX_DEPTH` → `DEFAULT_MCE_MAX_DEPTH` - `DEFAULT_MULTI_KUIPER_MIN_SAMPLES_PER_SEGMENT` → `DEFAULT_MCE_MIN_SAMPLES_PER_SEGMENT` - `DEFAULT_MULTI_KUIPER_GLOBAL_NORMALIZATION` → `DEFAULT_MCE_GLOBAL_NORMALIZATION` - `DEFAULT_MULTI_KUIPER_N_SEGMENTS` → `DEFAULT_MCE_N_SEGMENTS` Also updates fluent2 consumers to use the new constant names: - `fblearner/flow/projects/fluent2/definition/metrics/multicalibration_error.py` - `fblearner/flow/projects/fluent2/definition/evaluation.py` **Test Plan:** ``` buck2 test multicalibration/tests:test_metrics buck2 test fblearner/flow/projects/fluent2/workflows/evaluation/metrics/tests:multicalibration_error_test ``` Differential Revision: D91306073
flinder
force-pushed
the
export-D91306073
branch
2 times, most recently
from
b831112 to
a2e0a51
Compare
flinder
added a commit
to flinder/MCGrad-1
that referenced
this pull request
Jan 26, 2026
…T_MCE (facebookincubator#169) Summary: Renames module-level constants, some variables and docstrings from legacy Kuiper naming to MCE (Multicalibration Error) naming for consistency with the ECCE API refactoring: - `DEFAULT_MULTI_KUIPER_MAX_VALUES_PER_SEGMENT_FEATURE` → `DEFAULT_MCE_MAX_VALUES_PER_SEGMENT_FEATURE` - `DEFAULT_MULTI_KUIPER_MIN_DEPTH` → `DEFAULT_MCE_MIN_DEPTH` - `DEFAULT_MULTI_KUIPER_MAX_DEPTH` → `DEFAULT_MCE_MAX_DEPTH` - `DEFAULT_MULTI_KUIPER_MIN_SAMPLES_PER_SEGMENT` → `DEFAULT_MCE_MIN_SAMPLES_PER_SEGMENT` - `DEFAULT_MULTI_KUIPER_GLOBAL_NORMALIZATION` → `DEFAULT_MCE_GLOBAL_NORMALIZATION` - `DEFAULT_MULTI_KUIPER_N_SEGMENTS` → `DEFAULT_MCE_N_SEGMENTS` Also updates fluent2 consumers to use the new constant names: - `fblearner/flow/projects/fluent2/definition/metrics/multicalibration_error.py` - `fblearner/flow/projects/fluent2/definition/evaluation.py` **Test Plan:** ``` buck2 test multicalibration/tests:test_metrics buck2 test fblearner/flow/projects/fluent2/workflows/evaluation/metrics/tests:multicalibration_error_test ``` Reviewed By: Lorenzo-Perini Differential Revision: D91306073
flinder
added a commit
to flinder/MCGrad-1
that referenced
this pull request
Jan 26, 2026
…T_MCE (facebookincubator#169) Summary: Renames module-level constants, some variables and docstrings from legacy Kuiper naming to MCE (Multicalibration Error) naming for consistency with the ECCE API refactoring: - `DEFAULT_MULTI_KUIPER_MAX_VALUES_PER_SEGMENT_FEATURE` → `DEFAULT_MCE_MAX_VALUES_PER_SEGMENT_FEATURE` - `DEFAULT_MULTI_KUIPER_MIN_DEPTH` → `DEFAULT_MCE_MIN_DEPTH` - `DEFAULT_MULTI_KUIPER_MAX_DEPTH` → `DEFAULT_MCE_MAX_DEPTH` - `DEFAULT_MULTI_KUIPER_MIN_SAMPLES_PER_SEGMENT` → `DEFAULT_MCE_MIN_SAMPLES_PER_SEGMENT` - `DEFAULT_MULTI_KUIPER_GLOBAL_NORMALIZATION` → `DEFAULT_MCE_GLOBAL_NORMALIZATION` - `DEFAULT_MULTI_KUIPER_N_SEGMENTS` → `DEFAULT_MCE_N_SEGMENTS` Also updates fluent2 consumers to use the new constant names: - `fblearner/flow/projects/fluent2/definition/metrics/multicalibration_error.py` - `fblearner/flow/projects/fluent2/definition/evaluation.py` **Test Plan:** ``` buck2 test multicalibration/tests:test_metrics buck2 test fblearner/flow/projects/fluent2/workflows/evaluation/metrics/tests:multicalibration_error_test ``` Reviewed By: Lorenzo-Perini Differential Revision: D91306073
flinder
added a commit
to flinder/MCGrad-1
that referenced
this pull request
Jan 26, 2026
…T_MCE (facebookincubator#169) Summary: Pull Request resolved: facebookincubator#169 Renames module-level constants, some variables and docstrings from legacy Kuiper naming to MCE (Multicalibration Error) naming for consistency with the ECCE API refactoring: - `DEFAULT_MULTI_KUIPER_MAX_VALUES_PER_SEGMENT_FEATURE` → `DEFAULT_MCE_MAX_VALUES_PER_SEGMENT_FEATURE` - `DEFAULT_MULTI_KUIPER_MIN_DEPTH` → `DEFAULT_MCE_MIN_DEPTH` - `DEFAULT_MULTI_KUIPER_MAX_DEPTH` → `DEFAULT_MCE_MAX_DEPTH` - `DEFAULT_MULTI_KUIPER_MIN_SAMPLES_PER_SEGMENT` → `DEFAULT_MCE_MIN_SAMPLES_PER_SEGMENT` - `DEFAULT_MULTI_KUIPER_GLOBAL_NORMALIZATION` → `DEFAULT_MCE_GLOBAL_NORMALIZATION` - `DEFAULT_MULTI_KUIPER_N_SEGMENTS` → `DEFAULT_MCE_N_SEGMENTS` Also updates fluent2 consumers to use the new constant names: - `fblearner/flow/projects/fluent2/definition/metrics/multicalibration_error.py` - `fblearner/flow/projects/fluent2/definition/evaluation.py` **Test Plan:** ``` buck2 test multicalibration/tests:test_metrics buck2 test fblearner/flow/projects/fluent2/workflows/evaluation/metrics/tests:multicalibration_error_test ``` Reviewed By: Lorenzo-Perini Differential Revision: D91306073
flinder
force-pushed
the
export-D91306073
branch
from
a2e0a51 to
3d3ee69
Compare
flinder
added a commit
to flinder/MCGrad-1
that referenced
this pull request
Jan 26, 2026
…T_MCE (facebookincubator#169) Summary: Pull Request resolved: facebookincubator#169 Renames module-level constants, some variables and docstrings from legacy Kuiper naming to MCE (Multicalibration Error) naming for consistency with the ECCE API refactoring: - `DEFAULT_MULTI_KUIPER_MAX_VALUES_PER_SEGMENT_FEATURE` → `DEFAULT_MCE_MAX_VALUES_PER_SEGMENT_FEATURE` - `DEFAULT_MULTI_KUIPER_MIN_DEPTH` → `DEFAULT_MCE_MIN_DEPTH` - `DEFAULT_MULTI_KUIPER_MAX_DEPTH` → `DEFAULT_MCE_MAX_DEPTH` - `DEFAULT_MULTI_KUIPER_MIN_SAMPLES_PER_SEGMENT` → `DEFAULT_MCE_MIN_SAMPLES_PER_SEGMENT` - `DEFAULT_MULTI_KUIPER_GLOBAL_NORMALIZATION` → `DEFAULT_MCE_GLOBAL_NORMALIZATION` - `DEFAULT_MULTI_KUIPER_N_SEGMENTS` → `DEFAULT_MCE_N_SEGMENTS` Also updates fluent2 consumers to use the new constant names: - `fblearner/flow/projects/fluent2/definition/metrics/multicalibration_error.py` - `fblearner/flow/projects/fluent2/definition/evaluation.py` **Test Plan:** ``` buck2 test multicalibration/tests:test_metrics buck2 test fblearner/flow/projects/fluent2/workflows/evaluation/metrics/tests:multicalibration_error_test ``` Reviewed By: Lorenzo-Perini Differential Revision: D91306073
flinder
force-pushed
the
export-D91306073
branch
from
3d3ee69 to
539386d
Compare
flinder
added a commit
to flinder/MCGrad-1
that referenced
this pull request
Jan 26, 2026
…T_MCE (facebookincubator#169) Summary: Renames module-level constants, some variables and docstrings from legacy Kuiper naming to MCE (Multicalibration Error) naming for consistency with the ECCE API refactoring: - `DEFAULT_MULTI_KUIPER_MAX_VALUES_PER_SEGMENT_FEATURE` → `DEFAULT_MCE_MAX_VALUES_PER_SEGMENT_FEATURE` - `DEFAULT_MULTI_KUIPER_MIN_DEPTH` → `DEFAULT_MCE_MIN_DEPTH` - `DEFAULT_MULTI_KUIPER_MAX_DEPTH` → `DEFAULT_MCE_MAX_DEPTH` - `DEFAULT_MULTI_KUIPER_MIN_SAMPLES_PER_SEGMENT` → `DEFAULT_MCE_MIN_SAMPLES_PER_SEGMENT` - `DEFAULT_MULTI_KUIPER_GLOBAL_NORMALIZATION` → `DEFAULT_MCE_GLOBAL_NORMALIZATION` - `DEFAULT_MULTI_KUIPER_N_SEGMENTS` → `DEFAULT_MCE_N_SEGMENTS` Also updates fluent2 consumers to use the new constant names: - `fblearner/flow/projects/fluent2/definition/metrics/multicalibration_error.py` - `fblearner/flow/projects/fluent2/definition/evaluation.py` **Test Plan:** ``` buck2 test multicalibration/tests:test_metrics buck2 test fblearner/flow/projects/fluent2/workflows/evaluation/metrics/tests:multicalibration_error_test ``` Reviewed By: Lorenzo-Perini Differential Revision: D91306073
flinder
force-pushed
the
export-D91306073
branch
from
539386d to
a916f6d
Compare
flinder
added a commit
to flinder/MCGrad-1
that referenced
this pull request
Jan 26, 2026
…T_MCE (facebookincubator#169) Summary: Renames module-level constants, some variables and docstrings from legacy Kuiper naming to MCE (Multicalibration Error) naming for consistency with the ECCE API refactoring: - `DEFAULT_MULTI_KUIPER_MAX_VALUES_PER_SEGMENT_FEATURE` → `DEFAULT_MCE_MAX_VALUES_PER_SEGMENT_FEATURE` - `DEFAULT_MULTI_KUIPER_MIN_DEPTH` → `DEFAULT_MCE_MIN_DEPTH` - `DEFAULT_MULTI_KUIPER_MAX_DEPTH` → `DEFAULT_MCE_MAX_DEPTH` - `DEFAULT_MULTI_KUIPER_MIN_SAMPLES_PER_SEGMENT` → `DEFAULT_MCE_MIN_SAMPLES_PER_SEGMENT` - `DEFAULT_MULTI_KUIPER_GLOBAL_NORMALIZATION` → `DEFAULT_MCE_GLOBAL_NORMALIZATION` - `DEFAULT_MULTI_KUIPER_N_SEGMENTS` → `DEFAULT_MCE_N_SEGMENTS` Also updates fluent2 consumers to use the new constant names: - `fblearner/flow/projects/fluent2/definition/metrics/multicalibration_error.py` - `fblearner/flow/projects/fluent2/definition/evaluation.py` **Test Plan:** ``` buck2 test multicalibration/tests:test_metrics buck2 test fblearner/flow/projects/fluent2/workflows/evaluation/metrics/tests:multicalibration_error_test ``` Reviewed By: Lorenzo-Perini Differential Revision: D91306073
flinder
force-pushed
the
export-D91306073
branch
from
a916f6d to
35f61df
Compare
flinder
added a commit
to flinder/MCGrad-1
that referenced
this pull request
Jan 26, 2026
…T_MCE (facebookincubator#169) Summary: Renames module-level constants, some variables and docstrings from legacy Kuiper naming to MCE (Multicalibration Error) naming for consistency with the ECCE API refactoring: - `DEFAULT_MULTI_KUIPER_MAX_VALUES_PER_SEGMENT_FEATURE` → `DEFAULT_MCE_MAX_VALUES_PER_SEGMENT_FEATURE` - `DEFAULT_MULTI_KUIPER_MIN_DEPTH` → `DEFAULT_MCE_MIN_DEPTH` - `DEFAULT_MULTI_KUIPER_MAX_DEPTH` → `DEFAULT_MCE_MAX_DEPTH` - `DEFAULT_MULTI_KUIPER_MIN_SAMPLES_PER_SEGMENT` → `DEFAULT_MCE_MIN_SAMPLES_PER_SEGMENT` - `DEFAULT_MULTI_KUIPER_GLOBAL_NORMALIZATION` → `DEFAULT_MCE_GLOBAL_NORMALIZATION` - `DEFAULT_MULTI_KUIPER_N_SEGMENTS` → `DEFAULT_MCE_N_SEGMENTS` Also updates fluent2 consumers to use the new constant names: - `fblearner/flow/projects/fluent2/definition/metrics/multicalibration_error.py` - `fblearner/flow/projects/fluent2/definition/evaluation.py` **Test Plan:** ``` buck2 test multicalibration/tests:test_metrics buck2 test fblearner/flow/projects/fluent2/workflows/evaluation/metrics/tests:multicalibration_error_test ``` Reviewed By: Lorenzo-Perini Differential Revision: D91306073
flinder
added a commit
to flinder/MCGrad-1
that referenced
this pull request
Jan 26, 2026
…T_MCE (facebookincubator#169) Summary: Renames module-level constants, some variables and docstrings from legacy Kuiper naming to MCE (Multicalibration Error) naming for consistency with the ECCE API refactoring: - `DEFAULT_MULTI_KUIPER_MAX_VALUES_PER_SEGMENT_FEATURE` → `DEFAULT_MCE_MAX_VALUES_PER_SEGMENT_FEATURE` - `DEFAULT_MULTI_KUIPER_MIN_DEPTH` → `DEFAULT_MCE_MIN_DEPTH` - `DEFAULT_MULTI_KUIPER_MAX_DEPTH` → `DEFAULT_MCE_MAX_DEPTH` - `DEFAULT_MULTI_KUIPER_MIN_SAMPLES_PER_SEGMENT` → `DEFAULT_MCE_MIN_SAMPLES_PER_SEGMENT` - `DEFAULT_MULTI_KUIPER_GLOBAL_NORMALIZATION` → `DEFAULT_MCE_GLOBAL_NORMALIZATION` - `DEFAULT_MULTI_KUIPER_N_SEGMENTS` → `DEFAULT_MCE_N_SEGMENTS` Also updates fluent2 consumers to use the new constant names: - `fblearner/flow/projects/fluent2/definition/metrics/multicalibration_error.py` - `fblearner/flow/projects/fluent2/definition/evaluation.py` **Test Plan:** ``` buck2 test multicalibration/tests:test_metrics buck2 test fblearner/flow/projects/fluent2/workflows/evaluation/metrics/tests:multicalibration_error_test ``` Reviewed By: Lorenzo-Perini Differential Revision: D91306073
…lue (facebookincubator#160) Summary: Introduces a cleaner public API for Expected Cumulative Calibration Error (ECCE) metrics. The new ecce(), ecce_sigma(), and ecce_pvalue() functions provide simplified interfaces to the existing kuiper calibration functionality, making the API more intuitive for users who want to compute calibration errors without needing to understand the underlying Kuiper test implementation details. Deprecation of old versions of the functions will come in follow up diffs. ------------------------------------- Target API design ------------------------------------- ``` ┌─────────────────────────────────────────────────────────────────────────────────┐ │ ECCE API ARCHITECTURE │ ├─────────────────────────────────────────────────────────────────────────────────┤ │ │ │ ╔═══════════════════════════════════════════════════════════════════════════╗ │ │ ║ PUBLIC API (User-Facing) ║ │ │ ╠═══════════════════════════════════════════════════════════════════════════╣ │ │ ║ ║ │ │ ║ ┌─────────────┐ ┌──────────────┐ ┌──────────────────────────┐ ║ │ │ ║ │ ecce() │ │ ecce_sigma() │ │ ecce_pvalue() │ ║ │ │ ║ │ │ │ │ │ │ ║ │ │ ║ │ Unnormalized│ │ Normalized │ │ P-value for │ ║ │ │ ║ │ calibration │───▶│ by std dev │───▶│ calibration test │ ║ │ │ ║ │ error │ │ (sigma) │ │ │ ║ │ │ ║ └──────┬──────┘ └──────┬───────┘ └────────────┬─────────────┘ ║ │ │ ║ │ │ │ ║ │ │ ╚══════════╪══════════════════╪═════════════════════════╪══════════════════╝ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌──────────────────────────┐ │ │ │ │ │ ecce_pvalue_from_sigma()│ │ │ │ │ │ (public helper) │ │ │ │ │ └────────────┬─────────────┘ │ │ │ │ │ │ │ ┌──────────┴──────────────────┴────────────────────────┴───────────────────┐ │ │ │ PRIVATE HELPERS │ │ │ ├──────────────────────────────────────────────────────────────────────────┤ │ │ │ │ │ │ │ ┌────────────────────────────────┐ ┌─────────────────────────────┐ │ │ │ │ │ _calculate_cumulative_ │ │ _ecce_cdf() │ │ │ │ │ │ differences() │ │ │ │ │ │ │ │ │ │ CDF for Brownian motion │ │ │ │ │ │ Core computation of weighted │ │ range (for p-value calc) │ │ │ │ │ │ cumulative differences │ │ │ │ │ │ │ └───────────────┬────────────────┘ └─────────────────────────────┘ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌────────────────────────────────┐ ┌─────────────────────────────┐ │ │ │ │ │ _ecce_per_segment() │ │ _ecce_standard_deviation() │ │ │ │ │ │ │ │ │ │ │ │ │ │ Vectorized ECCE computation │ │ Wrapper → calls per_segment│ │ │ │ │ │ for multiple segments │ │ version with .item() │ │ │ │ │ └────────────────────────────────┘ └──────────────┬──────────────┘ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌─────────────────────────────┐ │ │ │ │ │ _ecce_standard_deviation_ │ │ │ │ │ │ per_segment() │ │ │ │ │ │ │ │ │ │ │ │ Variance-based std dev │ │ │ │ │ │ for multiple segments │ │ │ │ │ └─────────────────────────────┘ │ │ │ │ │ │ │ └──────────────────────────────────────────────────────────────────────────┘ │ │ │ │ ╔═══════════════════════════════════════════════════════════════════════════╗ │ │ ║ MulticalibrationError Class ║ │ │ ╠═══════════════════════════════════════════════════════════════════════════╣ │ │ ║ ║ │ │ ║ Computes calibration metrics across multiple data segments ║ │ │ ║ ║ │ │ ║ ┌─────────────────────────────────────────────────────────────────────┐ ║ │ │ ║ │ Cached Properties │ ║ │ │ ║ ├─────────────────────────────────────────────────────────────────────┤ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_ecces_absolute ──────────▶ _ecce_per_segment() │ ║ │ │ ║ │ (ECCE per segment) │ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_sigmas ──────────────────▶ _ecce_standard_deviation_ │ ║ │ │ ║ │ (std dev per segment) per_segment() │ ║ │ │ ║ │ │ ║ │ │ ║ │ sigma_0 ─────────────────────────▶ _ecce_standard_deviation_ │ ║ │ │ ║ │ (global std dev) per_segment() │ ║ │ │ ║ │ │ ║ │ │ ║ │ p_value ─────────────────────────▶ ecce_pvalue_from_sigma() │ ║ │ │ ║ │ (min p-value across segments) │ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_p_values ────────────────▶ ecce_pvalue_from_sigma() │ ║ │ │ ║ │ (p-value per segment) (vectorized) │ ║ │ │ ║ │ │ ║ │ │ ║ └─────────────────────────────────────────────────────────────────────┘ ║ │ │ ║ ║ │ │ ║ ┌─────────────────────────────────────────────────────────────────────┐ ║ │ │ ║ │ Derived Metrics │ ║ │ │ ║ ├─────────────────────────────────────────────────────────────────────┤ ║ │ │ ║ │ │ ║ │ │ ║ │ mce = max(segment_ecces_absolute) │ ║ │ │ ║ │ mce_sigma_scale = max(segment_ecces_absolute / segment_sigmas) │ ║ │ │ ║ │ mce_relative = mce / ecce_0 (normalized by global ECCE) │ ║ │ │ ║ │ │ ║ │ │ ║ └─────────────────────────────────────────────────────────────────────┘ ║ │ │ ║ ║ │ │ ╚═══════════════════════════════════════════════════════════════════════════╝ │ │ │ └─────────────────────────────────────────────────────────────────────────────────┘ ``` Reviewed By: TaXxER Differential Revision: D91235672
Summary: Removes the public `kuiper_calibration` function and updates internal callers to use the new ECCE API or lower-level functions directly. ### Changes **Removed:** - `kuiper_calibration()` - wrapper function that computed Kuiper calibration distance **Refactored:** - `kuiper_test()` - now calls `ecce_sigma()` directly instead of `kuiper_calibration(..., normalization_method="kuiper_standard_deviation")` [Note: this function will be removed in a future commit] - `ecce()` - now calls `kuiper_calibration_per_segment(...).item()` directly instead of going through `kuiper_calibration()` - `ecce_sigma()` - now calls `kuiper_calibration_per_segment(...).item()` directly instead of going through `kuiper_calibration()` ### Rationale This is part of a cleanup series migrating from the legacy "kuiper" naming to the cleaner "ECCE" (Expected Cumulative Calibration Error) API. The `kuiper_calibration` function was just a thin wrapper that called `kuiper_calibration_per_segment().item()`, so callers now use that pattern directly. ------------------------------------- Target API design ------------------------------------- ``` ┌─────────────────────────────────────────────────────────────────────────────────┐ │ ECCE API ARCHITECTURE │ ├─────────────────────────────────────────────────────────────────────────────────┤ │ │ │ ╔═══════════════════════════════════════════════════════════════════════════╗ │ │ ║ PUBLIC API (User-Facing) ║ │ │ ╠═══════════════════════════════════════════════════════════════════════════╣ │ │ ║ ║ │ │ ║ ┌─────────────┐ ┌──────────────┐ ┌──────────────────────────┐ ║ │ │ ║ │ ecce() │ │ ecce_sigma() │ │ ecce_pvalue() │ ║ │ │ ║ │ │ │ │ │ │ ║ │ │ ║ │ Unnormalized│ │ Normalized │ │ P-value for │ ║ │ │ ║ │ calibration │───▶│ by std dev │───▶│ calibration test │ ║ │ │ ║ │ error │ │ (sigma) │ │ │ ║ │ │ ║ └──────┬──────┘ └──────┬───────┘ └────────────┬─────────────┘ ║ │ │ ║ │ │ │ ║ │ │ ╚══════════╪══════════════════╪═════════════════════════╪══════════════════╝ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌──────────────────────────┐ │ │ │ │ │ ecce_pvalue_from_sigma()│ │ │ │ │ │ (public helper) │ │ │ │ │ └────────────┬─────────────┘ │ │ │ │ │ │ │ ┌──────────┴──────────────────┴────────────────────────┴───────────────────┐ │ │ │ PRIVATE HELPERS │ │ │ ├──────────────────────────────────────────────────────────────────────────┤ │ │ │ │ │ │ │ ┌────────────────────────────────┐ ┌─────────────────────────────┐ │ │ │ │ │ _calculate_cumulative_ │ │ _ecce_cdf() │ │ │ │ │ │ differences() │ │ │ │ │ │ │ │ │ │ CDF for Brownian motion │ │ │ │ │ │ Core computation of weighted │ │ range (for p-value calc) │ │ │ │ │ │ cumulative differences │ │ │ │ │ │ │ └───────────────┬────────────────┘ └─────────────────────────────┘ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌────────────────────────────────┐ ┌─────────────────────────────┐ │ │ │ │ │ _ecce_per_segment() │ │ _ecce_standard_deviation() │ │ │ │ │ │ │ │ │ │ │ │ │ │ Vectorized ECCE computation │ │ Wrapper → calls per_segment│ │ │ │ │ │ for multiple segments │ │ version with .item() │ │ │ │ │ └────────────────────────────────┘ └──────────────┬──────────────┘ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌─────────────────────────────┐ │ │ │ │ │ _ecce_standard_deviation_ │ │ │ │ │ │ per_segment() │ │ │ │ │ │ │ │ │ │ │ │ Variance-based std dev │ │ │ │ │ │ for multiple segments │ │ │ │ │ └─────────────────────────────┘ │ │ │ │ │ │ │ └──────────────────────────────────────────────────────────────────────────┘ │ │ │ │ ╔═══════════════════════════════════════════════════════════════════════════╗ │ │ ║ MulticalibrationError Class ║ │ │ ╠═══════════════════════════════════════════════════════════════════════════╣ │ │ ║ ║ │ │ ║ Computes calibration metrics across multiple data segments ║ │ │ ║ ║ │ │ ║ ┌─────────────────────────────────────────────────────────────────────┐ ║ │ │ ║ │ Cached Properties │ ║ │ │ ║ ├─────────────────────────────────────────────────────────────────────┤ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_ecces_absolute ──────────▶ _ecce_per_segment() │ ║ │ │ ║ │ (ECCE per segment) │ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_sigmas ──────────────────▶ _ecce_standard_deviation_ │ ║ │ │ ║ │ (std dev per segment) per_segment() │ ║ │ │ ║ │ │ ║ │ │ ║ │ sigma_0 ─────────────────────────▶ _ecce_standard_deviation_ │ ║ │ │ ║ │ (global std dev) per_segment() │ ║ │ │ ║ │ │ ║ │ │ ║ │ p_value ─────────────────────────▶ ecce_pvalue_from_sigma() │ ║ │ │ ║ │ (min p-value across segments) │ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_p_values ────────────────▶ ecce_pvalue_from_sigma() │ ║ │ │ ║ │ (p-value per segment) (vectorized) │ ║ │ │ ║ │ │ ║ │ │ ║ └─────────────────────────────────────────────────────────────────────┘ ║ │ │ ║ ║ │ │ ║ ┌─────────────────────────────────────────────────────────────────────┐ ║ │ │ ║ │ Derived Metrics │ ║ │ │ ║ ├─────────────────────────────────────────────────────────────────────┤ ║ │ │ ║ │ │ ║ │ │ ║ │ mce = max(segment_ecces_absolute) │ ║ │ │ ║ │ mce_sigma_scale = max(segment_ecces_absolute / segment_sigmas) │ ║ │ │ ║ │ mce_relative = mce / ecce_0 (normalized by global ECCE) │ ║ │ │ ║ │ │ ║ │ │ ║ └─────────────────────────────────────────────────────────────────────┘ ║ │ │ ║ ║ │ │ ╚═══════════════════════════════════════════════════════════════════════════╝ │ │ │ └─────────────────────────────────────────────────────────────────────────────────┘ ``` Reviewed By: Lorenzo-Perini Differential Revision: D91235674
) Summary: * Remove the function `kuiper_distribution` and replace with private helper `_ecce_cdf` * Add public helper `ecce_pvalue_from_sigma()` to compute p-value from sigma-scaled ECCE statistic, encapsulating the min/max checking logic * Rename constants `KUIPER_STATISTIC_MIN/MAX` to `ECCE_SIGMA_MIN/MAX` * Refactor all internal usages (`kuiper_test`, `ecce_pvalue`, `MulticalibrationError.p_value`, `MulticalibrationError.segment_p_values`) to use the new helpers * Update external call site in swisscheese `mce.py` to use `ecce_pvalue_from_sigma` ------------------------------------- Target API design ------------------------------------- ``` ┌─────────────────────────────────────────────────────────────────────────────────┐ │ ECCE API ARCHITECTURE │ ├─────────────────────────────────────────────────────────────────────────────────┤ │ │ │ ╔═══════════════════════════════════════════════════════════════════════════╗ │ │ ║ PUBLIC API (User-Facing) ║ │ │ ╠═══════════════════════════════════════════════════════════════════════════╣ │ │ ║ ║ │ │ ║ ┌─────────────┐ ┌──────────────┐ ┌──────────────────────────┐ ║ │ │ ║ │ ecce() │ │ ecce_sigma() │ │ ecce_pvalue() │ ║ │ │ ║ │ │ │ │ │ │ ║ │ │ ║ │ Unnormalized│ │ Normalized │ │ P-value for │ ║ │ │ ║ │ calibration │───▶│ by std dev │───▶│ calibration test │ ║ │ │ ║ │ error │ │ (sigma) │ │ │ ║ │ │ ║ └──────┬──────┘ └──────┬───────┘ └────────────┬─────────────┘ ║ │ │ ║ │ │ │ ║ │ │ ╚══════════╪══════════════════╪═════════════════════════╪══════════════════╝ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌──────────────────────────┐ │ │ │ │ │ ecce_pvalue_from_sigma()│ │ │ │ │ │ (public helper) │ │ │ │ │ └────────────┬─────────────┘ │ │ │ │ │ │ │ ┌──────────┴──────────────────┴────────────────────────┴───────────────────┐ │ │ │ PRIVATE HELPERS │ │ │ ├──────────────────────────────────────────────────────────────────────────┤ │ │ │ │ │ │ │ ┌────────────────────────────────┐ ┌─────────────────────────────┐ │ │ │ │ │ _calculate_cumulative_ │ │ _ecce_cdf() │ │ │ │ │ │ differences() │ │ │ │ │ │ │ │ │ │ CDF for Brownian motion │ │ │ │ │ │ Core computation of weighted │ │ range (for p-value calc) │ │ │ │ │ │ cumulative differences │ │ │ │ │ │ │ └───────────────┬────────────────┘ └─────────────────────────────┘ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌────────────────────────────────┐ ┌─────────────────────────────┐ │ │ │ │ │ _ecce_per_segment() │ │ _ecce_standard_deviation() │ │ │ │ │ │ │ │ │ │ │ │ │ │ Vectorized ECCE computation │ │ Wrapper → calls per_segment│ │ │ │ │ │ for multiple segments │ │ version with .item() │ │ │ │ │ └────────────────────────────────┘ └──────────────┬──────────────┘ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌─────────────────────────────┐ │ │ │ │ │ _ecce_standard_deviation_ │ │ │ │ │ │ per_segment() │ │ │ │ │ │ │ │ │ │ │ │ Variance-based std dev │ │ │ │ │ │ for multiple segments │ │ │ │ │ └─────────────────────────────┘ │ │ │ │ │ │ │ └──────────────────────────────────────────────────────────────────────────┘ │ │ │ │ ╔═══════════════════════════════════════════════════════════════════════════╗ │ │ ║ MulticalibrationError Class ║ │ │ ╠═══════════════════════════════════════════════════════════════════════════╣ │ │ ║ ║ │ │ ║ Computes calibration metrics across multiple data segments ║ │ │ ║ ║ │ │ ║ ┌─────────────────────────────────────────────────────────────────────┐ ║ │ │ ║ │ Cached Properties │ ║ │ │ ║ ├─────────────────────────────────────────────────────────────────────┤ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_ecces_absolute ──────────▶ _ecce_per_segment() │ ║ │ │ ║ │ (ECCE per segment) │ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_sigmas ──────────────────▶ _ecce_standard_deviation_ │ ║ │ │ ║ │ (std dev per segment) per_segment() │ ║ │ │ ║ │ │ ║ │ │ ║ │ sigma_0 ─────────────────────────▶ _ecce_standard_deviation_ │ ║ │ │ ║ │ (global std dev) per_segment() │ ║ │ │ ║ │ │ ║ │ │ ║ │ p_value ─────────────────────────▶ ecce_pvalue_from_sigma() │ ║ │ │ ║ │ (min p-value across segments) │ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_p_values ────────────────▶ ecce_pvalue_from_sigma() │ ║ │ │ ║ │ (p-value per segment) (vectorized) │ ║ │ │ ║ │ │ ║ │ │ ║ └─────────────────────────────────────────────────────────────────────┘ ║ │ │ ║ ║ │ │ ║ ┌─────────────────────────────────────────────────────────────────────┐ ║ │ │ ║ │ Derived Metrics │ ║ │ │ ║ ├─────────────────────────────────────────────────────────────────────┤ ║ │ │ ║ │ │ ║ │ │ ║ │ mce = max(segment_ecces_absolute) │ ║ │ │ ║ │ mce_sigma_scale = max(segment_ecces_absolute / segment_sigmas) │ ║ │ │ ║ │ mce_relative = mce / ecce_0 (normalized by global ECCE) │ ║ │ │ ║ │ │ ║ │ │ ║ └─────────────────────────────────────────────────────────────────────┘ ║ │ │ ║ ║ │ │ ╚═══════════════════════════════════════════════════════════════════════════╝ │ │ │ └─────────────────────────────────────────────────────────────────────────────────┘ ``` Reviewed By: Lorenzo-Perini Differential Revision: D91298776
Summary: Removes the deprecated kuiper_pvalue function and migrates all usages to the new ecce_pvalue API. ## Changes **Removed from metrics.py:** * kuiper_pvalue() function (callers should use ecce_pvalue() instead) **Updated external call sites**: * fbcode/fblearner/flow/projects/fluent2/workflows/evaluation/metrics/kuiper_calibration_pvalue.py: Renamed _compute_kuiper_pvalue() → _compute_ecce_pvalue() Now calls multicalibration_metrics.ecce_pvalue() instead of kuiper_pvalue() Updated tests: * test_metrics.py: Renamed test_kuiper_pvalue_wrapper_returns_pvalue_only → test_ecce_pvalue_wrapper_returns_pvalue_only and updated to use metrics.ecce_pvalue() ------------------------------------- Target API design ------------------------------------- ``` ┌─────────────────────────────────────────────────────────────────────────────────┐ │ ECCE API ARCHITECTURE │ ├─────────────────────────────────────────────────────────────────────────────────┤ │ │ │ ╔═══════════════════════════════════════════════════════════════════════════╗ │ │ ║ PUBLIC API (User-Facing) ║ │ │ ╠═══════════════════════════════════════════════════════════════════════════╣ │ │ ║ ║ │ │ ║ ┌─────────────┐ ┌──────────────┐ ┌──────────────────────────┐ ║ │ │ ║ │ ecce() │ │ ecce_sigma() │ │ ecce_pvalue() │ ║ │ │ ║ │ │ │ │ │ │ ║ │ │ ║ │ Unnormalized│ │ Normalized │ │ P-value for │ ║ │ │ ║ │ calibration │───▶│ by std dev │───▶│ calibration test │ ║ │ │ ║ │ error │ │ (sigma) │ │ │ ║ │ │ ║ └──────┬──────┘ └──────┬───────┘ └────────────┬─────────────┘ ║ │ │ ║ │ │ │ ║ │ │ ╚══════════╪══════════════════╪═════════════════════════╪══════════════════╝ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌──────────────────────────┐ │ │ │ │ │ ecce_pvalue_from_sigma()│ │ │ │ │ │ (public helper) │ │ │ │ │ └────────────┬─────────────┘ │ │ │ │ │ │ │ ┌──────────┴──────────────────┴────────────────────────┴───────────────────┐ │ │ │ PRIVATE HELPERS │ │ │ ├──────────────────────────────────────────────────────────────────────────┤ │ │ │ │ │ │ │ ┌────────────────────────────────┐ ┌─────────────────────────────┐ │ │ │ │ │ _calculate_cumulative_ │ │ _ecce_cdf() │ │ │ │ │ │ differences() │ │ │ │ │ │ │ │ │ │ CDF for Brownian motion │ │ │ │ │ │ Core computation of weighted │ │ range (for p-value calc) │ │ │ │ │ │ cumulative differences │ │ │ │ │ │ │ └───────────────┬────────────────┘ └─────────────────────────────┘ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌────────────────────────────────┐ ┌─────────────────────────────┐ │ │ │ │ │ _ecce_per_segment() │ │ _ecce_standard_deviation() │ │ │ │ │ │ │ │ │ │ │ │ │ │ Vectorized ECCE computation │ │ Wrapper → calls per_segment│ │ │ │ │ │ for multiple segments │ │ version with .item() │ │ │ │ │ └────────────────────────────────┘ └──────────────┬──────────────┘ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌─────────────────────────────┐ │ │ │ │ │ _ecce_standard_deviation_ │ │ │ │ │ │ per_segment() │ │ │ │ │ │ │ │ │ │ │ │ Variance-based std dev │ │ │ │ │ │ for multiple segments │ │ │ │ │ └─────────────────────────────┘ │ │ │ │ │ │ │ └──────────────────────────────────────────────────────────────────────────┘ │ │ │ │ ╔═══════════════════════════════════════════════════════════════════════════╗ │ │ ║ MulticalibrationError Class ║ │ │ ╠═══════════════════════════════════════════════════════════════════════════╣ │ │ ║ ║ │ │ ║ Computes calibration metrics across multiple data segments ║ │ │ ║ ║ │ │ ║ ┌─────────────────────────────────────────────────────────────────────┐ ║ │ │ ║ │ Cached Properties │ ║ │ │ ║ ├─────────────────────────────────────────────────────────────────────┤ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_ecces_absolute ──────────▶ _ecce_per_segment() │ ║ │ │ ║ │ (ECCE per segment) │ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_sigmas ──────────────────▶ _ecce_standard_deviation_ │ ║ │ │ ║ │ (std dev per segment) per_segment() │ ║ │ │ ║ │ │ ║ │ │ ║ │ sigma_0 ─────────────────────────▶ _ecce_standard_deviation_ │ ║ │ │ ║ │ (global std dev) per_segment() │ ║ │ │ ║ │ │ ║ │ │ ║ │ p_value ─────────────────────────▶ ecce_pvalue_from_sigma() │ ║ │ │ ║ │ (min p-value across segments) │ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_p_values ────────────────▶ ecce_pvalue_from_sigma() │ ║ │ │ ║ │ (p-value per segment) (vectorized) │ ║ │ │ ║ │ │ ║ │ │ ║ └─────────────────────────────────────────────────────────────────────┘ ║ │ │ ║ ║ │ │ ║ ┌─────────────────────────────────────────────────────────────────────┐ ║ │ │ ║ │ Derived Metrics │ ║ │ │ ║ ├─────────────────────────────────────────────────────────────────────┤ ║ │ │ ║ │ │ ║ │ │ ║ │ mce = max(segment_ecces_absolute) │ ║ │ │ ║ │ mce_sigma_scale = max(segment_ecces_absolute / segment_sigmas) │ ║ │ │ ║ │ mce_relative = mce / ecce_0 (normalized by global ECCE) │ ║ │ │ ║ │ │ ║ │ │ ║ └─────────────────────────────────────────────────────────────────────┘ ║ │ │ ║ ║ │ │ ╚═══════════════════════════════════════════════════════════════════════════╝ │ │ │ └─────────────────────────────────────────────────────────────────────────────────┘ ``` Reviewed By: Lorenzo-Perini Differential Revision: D91298790
…kincubator#164) Summary: Removes the public `kuiper_calibration_per_segment` function and replaces it with a simplified private helper `_ecce_per_segment` that only computes unnormalized ECCE. Changes: - Replace `kuiper_calibration_per_segment` with `_ecce_per_segment` (removed normalization parameter since it was always None in the only call site) - Make `ecce()` implement its own logic directly using `_calculate_cumulative_differences` - Make `ecce_sigma()` call `ecce()` and divide by `kuiper_standard_deviation` with proper zero-handling - Update `MulticalibrationError.segment_ecces_absolute` to use the new `_ecce_per_segment` helper ------------------------------------- Target API design ------------------------------------- ``` ┌─────────────────────────────────────────────────────────────────────────────────┐ │ ECCE API ARCHITECTURE │ ├─────────────────────────────────────────────────────────────────────────────────┤ │ │ │ ╔═══════════════════════════════════════════════════════════════════════════╗ │ │ ║ PUBLIC API (User-Facing) ║ │ │ ╠═══════════════════════════════════════════════════════════════════════════╣ │ │ ║ ║ │ │ ║ ┌─────────────┐ ┌──────────────┐ ┌──────────────────────────┐ ║ │ │ ║ │ ecce() │ │ ecce_sigma() │ │ ecce_pvalue() │ ║ │ │ ║ │ │ │ │ │ │ ║ │ │ ║ │ Unnormalized│ │ Normalized │ │ P-value for │ ║ │ │ ║ │ calibration │───▶│ by std dev │───▶│ calibration test │ ║ │ │ ║ │ error │ │ (sigma) │ │ │ ║ │ │ ║ └──────┬──────┘ └──────┬───────┘ └────────────┬─────────────┘ ║ │ │ ║ │ │ │ ║ │ │ ╚══════════╪══════════════════╪═════════════════════════╪══════════════════╝ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌──────────────────────────┐ │ │ │ │ │ ecce_pvalue_from_sigma()│ │ │ │ │ │ (public helper) │ │ │ │ │ └────────────┬─────────────┘ │ │ │ │ │ │ │ ┌──────────┴──────────────────┴────────────────────────┴───────────────────┐ │ │ │ PRIVATE HELPERS │ │ │ ├──────────────────────────────────────────────────────────────────────────┤ │ │ │ │ │ │ │ ┌────────────────────────────────┐ ┌─────────────────────────────┐ │ │ │ │ │ _calculate_cumulative_ │ │ _ecce_cdf() │ │ │ │ │ │ differences() │ │ │ │ │ │ │ │ │ │ CDF for Brownian motion │ │ │ │ │ │ Core computation of weighted │ │ range (for p-value calc) │ │ │ │ │ │ cumulative differences │ │ │ │ │ │ │ └───────────────┬────────────────┘ └─────────────────────────────┘ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌────────────────────────────────┐ ┌─────────────────────────────┐ │ │ │ │ │ _ecce_per_segment() │ │ _ecce_standard_deviation() │ │ │ │ │ │ │ │ │ │ │ │ │ │ Vectorized ECCE computation │ │ Wrapper → calls per_segment│ │ │ │ │ │ for multiple segments │ │ version with .item() │ │ │ │ │ └────────────────────────────────┘ └──────────────┬──────────────┘ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌─────────────────────────────┐ │ │ │ │ │ _ecce_standard_deviation_ │ │ │ │ │ │ per_segment() │ │ │ │ │ │ │ │ │ │ │ │ Variance-based std dev │ │ │ │ │ │ for multiple segments │ │ │ │ │ └─────────────────────────────┘ │ │ │ │ │ │ │ └──────────────────────────────────────────────────────────────────────────┘ │ │ │ │ ╔═══════════════════════════════════════════════════════════════════════════╗ │ │ ║ MulticalibrationError Class ║ │ │ ╠═══════════════════════════════════════════════════════════════════════════╣ │ │ ║ ║ │ │ ║ Computes calibration metrics across multiple data segments ║ │ │ ║ ║ │ │ ║ ┌─────────────────────────────────────────────────────────────────────┐ ║ │ │ ║ │ Cached Properties │ ║ │ │ ║ ├─────────────────────────────────────────────────────────────────────┤ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_ecces_absolute ──────────▶ _ecce_per_segment() │ ║ │ │ ║ │ (ECCE per segment) │ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_sigmas ──────────────────▶ _ecce_standard_deviation_ │ ║ │ │ ║ │ (std dev per segment) per_segment() │ ║ │ │ ║ │ │ ║ │ │ ║ │ sigma_0 ─────────────────────────▶ _ecce_standard_deviation_ │ ║ │ │ ║ │ (global std dev) per_segment() │ ║ │ │ ║ │ │ ║ │ │ ║ │ p_value ─────────────────────────▶ ecce_pvalue_from_sigma() │ ║ │ │ ║ │ (min p-value across segments) │ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_p_values ────────────────▶ ecce_pvalue_from_sigma() │ ║ │ │ ║ │ (p-value per segment) (vectorized) │ ║ │ │ ║ │ │ ║ │ │ ║ └─────────────────────────────────────────────────────────────────────┘ ║ │ │ ║ ║ │ │ ║ ┌─────────────────────────────────────────────────────────────────────┐ ║ │ │ ║ │ Derived Metrics │ ║ │ │ ║ ├─────────────────────────────────────────────────────────────────────┤ ║ │ │ ║ │ │ ║ │ │ ║ │ mce = max(segment_ecces_absolute) │ ║ │ │ ║ │ mce_sigma_scale = max(segment_ecces_absolute / segment_sigmas) │ ║ │ │ ║ │ mce_relative = mce / ecce_0 (normalized by global ECCE) │ ║ │ │ ║ │ │ ║ │ │ ║ └─────────────────────────────────────────────────────────────────────┘ ║ │ │ ║ ║ │ │ ╚═══════════════════════════════════════════════════════════════════════════╝ │ │ │ └─────────────────────────────────────────────────────────────────────────────────┘ ``` Reviewed By: Lorenzo-Perini Differential Revision: D91299869
…ubator#168) Summary: Makes the `kuiper_standard_deviation` and `kuiper_standard_deviation_per_segment` functions private helpers with ECCE naming: - Rename `kuiper_standard_deviation` → `_ecce_standard_deviation` - Rename `kuiper_standard_deviation_per_segment` → `_ecce_standard_deviation_per_segment` - Update all internal usages in `ecce_sigma()` and tests - The `_normalization_method_assignment` mapping still supports the string `"kuiper_standard_deviation"` for backward compatibility ------------------------------------- Target API design ------------------------------------- ``` ┌─────────────────────────────────────────────────────────────────────────────────┐ │ ECCE API ARCHITECTURE │ ├─────────────────────────────────────────────────────────────────────────────────┤ │ │ │ ╔═══════════════════════════════════════════════════════════════════════════╗ │ │ ║ PUBLIC API (User-Facing) ║ │ │ ╠═══════════════════════════════════════════════════════════════════════════╣ │ │ ║ ║ │ │ ║ ┌─────────────┐ ┌──────────────┐ ┌──────────────────────────┐ ║ │ │ ║ │ ecce() │ │ ecce_sigma() │ │ ecce_pvalue() │ ║ │ │ ║ │ │ │ │ │ │ ║ │ │ ║ │ Unnormalized│ │ Normalized │ │ P-value for │ ║ │ │ ║ │ calibration │───▶│ by std dev │───▶│ calibration test │ ║ │ │ ║ │ error │ │ (sigma) │ │ │ ║ │ │ ║ └──────┬──────┘ └──────┬───────┘ └────────────┬─────────────┘ ║ │ │ ║ │ │ │ ║ │ │ ╚══════════╪══════════════════╪═════════════════════════╪══════════════════╝ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌──────────────────────────┐ │ │ │ │ │ ecce_pvalue_from_sigma()│ │ │ │ │ │ (public helper) │ │ │ │ │ └────────────┬─────────────┘ │ │ │ │ │ │ │ ┌──────────┴──────────────────┴────────────────────────┴───────────────────┐ │ │ │ PRIVATE HELPERS │ │ │ ├──────────────────────────────────────────────────────────────────────────┤ │ │ │ │ │ │ │ ┌────────────────────────────────┐ ┌─────────────────────────────┐ │ │ │ │ │ _calculate_cumulative_ │ │ _ecce_cdf() │ │ │ │ │ │ differences() │ │ │ │ │ │ │ │ │ │ CDF for Brownian motion │ │ │ │ │ │ Core computation of weighted │ │ range (for p-value calc) │ │ │ │ │ │ cumulative differences │ │ │ │ │ │ │ └───────────────┬────────────────┘ └─────────────────────────────┘ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌────────────────────────────────┐ ┌─────────────────────────────┐ │ │ │ │ │ _ecce_per_segment() │ │ _ecce_standard_deviation() │ │ │ │ │ │ │ │ │ │ │ │ │ │ Vectorized ECCE computation │ │ Wrapper → calls per_segment│ │ │ │ │ │ for multiple segments │ │ version with .item() │ │ │ │ │ └────────────────────────────────┘ └──────────────┬──────────────┘ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌─────────────────────────────┐ │ │ │ │ │ _ecce_standard_deviation_ │ │ │ │ │ │ per_segment() │ │ │ │ │ │ │ │ │ │ │ │ Variance-based std dev │ │ │ │ │ │ for multiple segments │ │ │ │ │ └─────────────────────────────┘ │ │ │ │ │ │ │ └──────────────────────────────────────────────────────────────────────────┘ │ │ │ │ ╔═══════════════════════════════════════════════════════════════════════════╗ │ │ ║ MulticalibrationError Class ║ │ │ ╠═══════════════════════════════════════════════════════════════════════════╣ │ │ ║ ║ │ │ ║ Computes calibration metrics across multiple data segments ║ │ │ ║ ║ │ │ ║ ┌─────────────────────────────────────────────────────────────────────┐ ║ │ │ ║ │ Cached Properties │ ║ │ │ ║ ├─────────────────────────────────────────────────────────────────────┤ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_ecces_absolute ──────────▶ _ecce_per_segment() │ ║ │ │ ║ │ (ECCE per segment) │ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_sigmas ──────────────────▶ _ecce_standard_deviation_ │ ║ │ │ ║ │ (std dev per segment) per_segment() │ ║ │ │ ║ │ │ ║ │ │ ║ │ sigma_0 ─────────────────────────▶ _ecce_standard_deviation_ │ ║ │ │ ║ │ (global std dev) per_segment() │ ║ │ │ ║ │ │ ║ │ │ ║ │ p_value ─────────────────────────▶ ecce_pvalue_from_sigma() │ ║ │ │ ║ │ (min p-value across segments) │ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_p_values ────────────────▶ ecce_pvalue_from_sigma() │ ║ │ │ ║ │ (p-value per segment) (vectorized) │ ║ │ │ ║ │ │ ║ │ │ ║ └─────────────────────────────────────────────────────────────────────┘ ║ │ │ ║ ║ │ │ ║ ┌─────────────────────────────────────────────────────────────────────┐ ║ │ │ ║ │ Derived Metrics │ ║ │ │ ║ ├─────────────────────────────────────────────────────────────────────┤ ║ │ │ ║ │ │ ║ │ │ ║ │ mce = max(segment_ecces_absolute) │ ║ │ │ ║ │ mce_sigma_scale = max(segment_ecces_absolute / segment_sigmas) │ ║ │ │ ║ │ mce_relative = mce / ecce_0 (normalized by global ECCE) │ ║ │ │ ║ │ │ ║ │ │ ║ └─────────────────────────────────────────────────────────────────────┘ ║ │ │ ║ ║ │ │ ╚═══════════════════════════════════════════════════════════════════════════╝ │ │ │ └─────────────────────────────────────────────────────────────────────────────────┘ ``` Reviewed By: Lorenzo-Perini Differential Revision: D91301861
…r#167) Summary: * Simplifies MulticalibrationError by removing the configurable sigma estimation method abstraction. We will only support ECCE standard deviation normalization, so the indirection layer is no longer needed. The class now uses `_ecce_standard_deviation_per_segment` directly instead of routing through a configurable `estimate_sigma` method. * Removes the sigma estimation method parameter from the fluent2 MCE workflow * Also removes all unused alternative standardization methods. ------------------------------------- Target API design ------------------------------------- ``` ┌─────────────────────────────────────────────────────────────────────────────────┐ │ ECCE API ARCHITECTURE │ ├─────────────────────────────────────────────────────────────────────────────────┤ │ │ │ ╔═══════════════════════════════════════════════════════════════════════════╗ │ │ ║ PUBLIC API (User-Facing) ║ │ │ ╠═══════════════════════════════════════════════════════════════════════════╣ │ │ ║ ║ │ │ ║ ┌─────────────┐ ┌──────────────┐ ┌──────────────────────────┐ ║ │ │ ║ │ ecce() │ │ ecce_sigma() │ │ ecce_pvalue() │ ║ │ │ ║ │ │ │ │ │ │ ║ │ │ ║ │ Unnormalized│ │ Normalized │ │ P-value for │ ║ │ │ ║ │ calibration │───▶│ by std dev │───▶│ calibration test │ ║ │ │ ║ │ error │ │ (sigma) │ │ │ ║ │ │ ║ └──────┬──────┘ └──────┬───────┘ └────────────┬─────────────┘ ║ │ │ ║ │ │ │ ║ │ │ ╚══════════╪══════════════════╪═════════════════════════╪══════════════════╝ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌──────────────────────────┐ │ │ │ │ │ ecce_pvalue_from_sigma()│ │ │ │ │ │ (public helper) │ │ │ │ │ └────────────┬─────────────┘ │ │ │ │ │ │ │ ┌──────────┴──────────────────┴────────────────────────┴───────────────────┐ │ │ │ PRIVATE HELPERS │ │ │ ├──────────────────────────────────────────────────────────────────────────┤ │ │ │ │ │ │ │ ┌────────────────────────────────┐ ┌─────────────────────────────┐ │ │ │ │ │ _calculate_cumulative_ │ │ _ecce_cdf() │ │ │ │ │ │ differences() │ │ │ │ │ │ │ │ │ │ CDF for Brownian motion │ │ │ │ │ │ Core computation of weighted │ │ range (for p-value calc) │ │ │ │ │ │ cumulative differences │ │ │ │ │ │ │ └───────────────┬────────────────┘ └─────────────────────────────┘ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌────────────────────────────────┐ ┌─────────────────────────────┐ │ │ │ │ │ _ecce_per_segment() │ │ _ecce_standard_deviation() │ │ │ │ │ │ │ │ │ │ │ │ │ │ Vectorized ECCE computation │ │ Wrapper → calls per_segment│ │ │ │ │ │ for multiple segments │ │ version with .item() │ │ │ │ │ └────────────────────────────────┘ └──────────────┬──────────────┘ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌─────────────────────────────┐ │ │ │ │ │ _ecce_standard_deviation_ │ │ │ │ │ │ per_segment() │ │ │ │ │ │ │ │ │ │ │ │ Variance-based std dev │ │ │ │ │ │ for multiple segments │ │ │ │ │ └─────────────────────────────┘ │ │ │ │ │ │ │ └──────────────────────────────────────────────────────────────────────────┘ │ │ │ │ ╔═══════════════════════════════════════════════════════════════════════════╗ │ │ ║ MulticalibrationError Class ║ │ │ ╠═══════════════════════════════════════════════════════════════════════════╣ │ │ ║ ║ │ │ ║ Computes calibration metrics across multiple data segments ║ │ │ ║ ║ │ │ ║ ┌─────────────────────────────────────────────────────────────────────┐ ║ │ │ ║ │ Cached Properties │ ║ │ │ ║ ├─────────────────────────────────────────────────────────────────────┤ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_ecces_absolute ──────────▶ _ecce_per_segment() │ ║ │ │ ║ │ (ECCE per segment) │ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_sigmas ──────────────────▶ _ecce_standard_deviation_ │ ║ │ │ ║ │ (std dev per segment) per_segment() │ ║ │ │ ║ │ │ ║ │ │ ║ │ sigma_0 ─────────────────────────▶ _ecce_standard_deviation_ │ ║ │ │ ║ │ (global std dev) per_segment() │ ║ │ │ ║ │ │ ║ │ │ ║ │ p_value ─────────────────────────▶ ecce_pvalue_from_sigma() │ ║ │ │ ║ │ (min p-value across segments) │ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_p_values ────────────────▶ ecce_pvalue_from_sigma() │ ║ │ │ ║ │ (p-value per segment) (vectorized) │ ║ │ │ ║ │ │ ║ │ │ ║ └─────────────────────────────────────────────────────────────────────┘ ║ │ │ ║ ║ │ │ ║ ┌─────────────────────────────────────────────────────────────────────┐ ║ │ │ ║ │ Derived Metrics │ ║ │ │ ║ ├─────────────────────────────────────────────────────────────────────┤ ║ │ │ ║ │ │ ║ │ │ ║ │ mce = max(segment_ecces_absolute) │ ║ │ │ ║ │ mce_sigma_scale = max(segment_ecces_absolute / segment_sigmas) │ ║ │ │ ║ │ mce_relative = mce / ecce_0 (normalized by global ECCE) │ ║ │ │ ║ │ │ ║ │ │ ║ └─────────────────────────────────────────────────────────────────────┘ ║ │ │ ║ ║ │ │ ╚═══════════════════════════════════════════════════════════════════════════╝ │ │ │ └─────────────────────────────────────────────────────────────────────────────────┘ ``` Reviewed By: Lorenzo-Perini Differential Revision: D91302327
… from metrics.py (facebookincubator#166) Summary: Changes: metrics.py Removes several deprecated/unused kuiper-related functions: kuiper_upper_bound_standard_deviation() kuiper_upper_bound_standard_deviation_per_segment() kuiper_label_based_standard_deviation() kuiper_label_based_standard_deviation_per_segment() _normalization_method_assignment() _KuiperNormalizationInterface type alias identity_per_segment() Also removes unused import Callable from typing. test_metrics.py Removes tests for the deleted functions: test_equivalence_kuiper_label_based_standard_deviation_per_segment test_kuiper_label_based_standard_deviation_edge_cases test_kuiper_label_based_standard_deviation_perfect_calibration test_kuiper_label_based_standard_deviation_requires_labels test_kuiper_upper_bound_standard_deviation_per_segment_with_default_sample_weight test_kuiper_upper_bound_standard_deviation_per_segment_with_default_segments test_kuiper_upper_bound_standard_deviation_wrapper_gives_expected_result test_kuiper_upper_bound_standard_deviation_with_sample_weight test_kuiper_label_based_standard_deviation_per_segment_with_default_sample_weight test_kuiper_label_based_standard_deviation_per_segment_with_default_segments test_kuiper_label_based_standard_deviation_per_segment_raises_error_when_labels_none test_normalization_method_assignment_raises_error_for_unknown_method This is part of a cleanup series simplifying the ECCE API by removing legacy kuiper naming and alternative standardization methods that are no longer used. ------------------------------------- Target API design ------------------------------------- ``` ┌─────────────────────────────────────────────────────────────────────────────────┐ │ ECCE API ARCHITECTURE │ ├─────────────────────────────────────────────────────────────────────────────────┤ │ │ │ ╔═══════════════════════════════════════════════════════════════════════════╗ │ │ ║ PUBLIC API (User-Facing) ║ │ │ ╠═══════════════════════════════════════════════════════════════════════════╣ │ │ ║ ║ │ │ ║ ┌─────────────┐ ┌──────────────┐ ┌──────────────────────────┐ ║ │ │ ║ │ ecce() │ │ ecce_sigma() │ │ ecce_pvalue() │ ║ │ │ ║ │ │ │ │ │ │ ║ │ │ ║ │ Unnormalized│ │ Normalized │ │ P-value for │ ║ │ │ ║ │ calibration │───▶│ by std dev │───▶│ calibration test │ ║ │ │ ║ │ error │ │ (sigma) │ │ │ ║ │ │ ║ └──────┬──────┘ └──────┬───────┘ └────────────┬─────────────┘ ║ │ │ ║ │ │ │ ║ │ │ ╚══════════╪══════════════════╪═════════════════════════╪══════════════════╝ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌──────────────────────────┐ │ │ │ │ │ ecce_pvalue_from_sigma()│ │ │ │ │ │ (public helper) │ │ │ │ │ └────────────┬─────────────┘ │ │ │ │ │ │ │ ┌──────────┴──────────────────┴────────────────────────┴───────────────────┐ │ │ │ PRIVATE HELPERS │ │ │ ├──────────────────────────────────────────────────────────────────────────┤ │ │ │ │ │ │ │ ┌────────────────────────────────┐ ┌─────────────────────────────┐ │ │ │ │ │ _calculate_cumulative_ │ │ _ecce_cdf() │ │ │ │ │ │ differences() │ │ │ │ │ │ │ │ │ │ CDF for Brownian motion │ │ │ │ │ │ Core computation of weighted │ │ range (for p-value calc) │ │ │ │ │ │ cumulative differences │ │ │ │ │ │ │ └───────────────┬────────────────┘ └─────────────────────────────┘ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌────────────────────────────────┐ ┌─────────────────────────────┐ │ │ │ │ │ _ecce_per_segment() │ │ _ecce_standard_deviation() │ │ │ │ │ │ │ │ │ │ │ │ │ │ Vectorized ECCE computation │ │ Wrapper → calls per_segment│ │ │ │ │ │ for multiple segments │ │ version with .item() │ │ │ │ │ └────────────────────────────────┘ └──────────────┬──────────────┘ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌─────────────────────────────┐ │ │ │ │ │ _ecce_standard_deviation_ │ │ │ │ │ │ per_segment() │ │ │ │ │ │ │ │ │ │ │ │ Variance-based std dev │ │ │ │ │ │ for multiple segments │ │ │ │ │ └─────────────────────────────┘ │ │ │ │ │ │ │ └──────────────────────────────────────────────────────────────────────────┘ │ │ │ │ ╔═══════════════════════════════════════════════════════════════════════════╗ │ │ ║ MulticalibrationError Class ║ │ │ ╠═══════════════════════════════════════════════════════════════════════════╣ │ │ ║ ║ │ │ ║ Computes calibration metrics across multiple data segments ║ │ │ ║ ║ │ │ ║ ┌─────────────────────────────────────────────────────────────────────┐ ║ │ │ ║ │ Cached Properties │ ║ │ │ ║ ├─────────────────────────────────────────────────────────────────────┤ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_ecces_absolute ──────────▶ _ecce_per_segment() │ ║ │ │ ║ │ (ECCE per segment) │ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_sigmas ──────────────────▶ _ecce_standard_deviation_ │ ║ │ │ ║ │ (std dev per segment) per_segment() │ ║ │ │ ║ │ │ ║ │ │ ║ │ sigma_0 ─────────────────────────▶ _ecce_standard_deviation_ │ ║ │ │ ║ │ (global std dev) per_segment() │ ║ │ │ ║ │ │ ║ │ │ ║ │ p_value ─────────────────────────▶ ecce_pvalue_from_sigma() │ ║ │ │ ║ │ (min p-value across segments) │ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_p_values ────────────────▶ ecce_pvalue_from_sigma() │ ║ │ │ ║ │ (p-value per segment) (vectorized) │ ║ │ │ ║ │ │ ║ │ │ ║ └─────────────────────────────────────────────────────────────────────┘ ║ │ │ ║ ║ │ │ ║ ┌─────────────────────────────────────────────────────────────────────┐ ║ │ │ ║ │ Derived Metrics │ ║ │ │ ║ ├─────────────────────────────────────────────────────────────────────┤ ║ │ │ ║ │ │ ║ │ │ ║ │ mce = max(segment_ecces_absolute) │ ║ │ │ ║ │ mce_sigma_scale = max(segment_ecces_absolute / segment_sigmas) │ ║ │ │ ║ │ mce_relative = mce / ecce_0 (normalized by global ECCE) │ ║ │ │ ║ │ │ ║ │ │ ║ └─────────────────────────────────────────────────────────────────────┘ ║ │ │ ║ ║ │ │ ╚═══════════════════════════════════════════════════════════════════════════╝ │ │ │ └─────────────────────────────────────────────────────────────────────────────────┘ ``` Reviewed By: Lorenzo-Perini Differential Revision: D91302516
Summary: Removes the `kuiper_test` function which was a legacy wrapper around `ecce_sigma` and `ecce_pvalue_from_sigma`. Users should call the ECCE functions directly instead. Changes: - Remove `kuiper_test()` function from `metrics.py` - Remove unused `scipy.stats` import - Update tests to use `ecce_pvalue()` and `ecce_sigma()` directly: - Rename `test_that_kuiper_test_detects_miscalibration` → `test_that_ecce_miscalibration` - Rename `test_kuiper_test_returns_pvalue_one_for_very_small_statistic` → `test_ecce_pvalue_returns_one_for_very_small_statistic` - Rename `test_kuiper_test_returns_epsilon_pvalue_for_very_large_statistic` → `test_ecce_pvalue_returns_epsilon_pvalue_for_very_large_statistic` - Rename `test_ecce_pvalue_consistency_with_kuiper_pvalue` → `test_ecce_pvalue_consistency_with_ecce_pvalue_from_sigma` ------------------------------------- Target API design ------------------------------------- ``` ┌─────────────────────────────────────────────────────────────────────────────────┐ │ ECCE API ARCHITECTURE │ ├─────────────────────────────────────────────────────────────────────────────────┤ │ │ │ ╔═══════════════════════════════════════════════════════════════════════════╗ │ │ ║ PUBLIC API (User-Facing) ║ │ │ ╠═══════════════════════════════════════════════════════════════════════════╣ │ │ ║ ║ │ │ ║ ┌─────────────┐ ┌──────────────┐ ┌──────────────────────────┐ ║ │ │ ║ │ ecce() │ │ ecce_sigma() │ │ ecce_pvalue() │ ║ │ │ ║ │ │ │ │ │ │ ║ │ │ ║ │ Unnormalized│ │ Normalized │ │ P-value for │ ║ │ │ ║ │ calibration │───▶│ by std dev │───▶│ calibration test │ ║ │ │ ║ │ error │ │ (sigma) │ │ │ ║ │ │ ║ └──────┬──────┘ └──────┬───────┘ └────────────┬─────────────┘ ║ │ │ ║ │ │ │ ║ │ │ ╚══════════╪══════════════════╪═════════════════════════╪══════════════════╝ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌──────────────────────────┐ │ │ │ │ │ ecce_pvalue_from_sigma()│ │ │ │ │ │ (public helper) │ │ │ │ │ └────────────┬─────────────┘ │ │ │ │ │ │ │ ┌──────────┴──────────────────┴────────────────────────┴───────────────────┐ │ │ │ PRIVATE HELPERS │ │ │ ├──────────────────────────────────────────────────────────────────────────┤ │ │ │ │ │ │ │ ┌────────────────────────────────┐ ┌─────────────────────────────┐ │ │ │ │ │ _calculate_cumulative_ │ │ _ecce_cdf() │ │ │ │ │ │ differences() │ │ │ │ │ │ │ │ │ │ CDF for Brownian motion │ │ │ │ │ │ Core computation of weighted │ │ range (for p-value calc) │ │ │ │ │ │ cumulative differences │ │ │ │ │ │ │ └───────────────┬────────────────┘ └─────────────────────────────┘ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌────────────────────────────────┐ ┌─────────────────────────────┐ │ │ │ │ │ _ecce_per_segment() │ │ _ecce_standard_deviation() │ │ │ │ │ │ │ │ │ │ │ │ │ │ Vectorized ECCE computation │ │ Wrapper → calls per_segment│ │ │ │ │ │ for multiple segments │ │ version with .item() │ │ │ │ │ └────────────────────────────────┘ └──────────────┬──────────────┘ │ │ │ │ │ │ │ │ │ ▼ │ │ │ │ ┌─────────────────────────────┐ │ │ │ │ │ _ecce_standard_deviation_ │ │ │ │ │ │ per_segment() │ │ │ │ │ │ │ │ │ │ │ │ Variance-based std dev │ │ │ │ │ │ for multiple segments │ │ │ │ │ └─────────────────────────────┘ │ │ │ │ │ │ │ └──────────────────────────────────────────────────────────────────────────┘ │ │ │ │ ╔═══════════════════════════════════════════════════════════════════════════╗ │ │ ║ MulticalibrationError Class ║ │ │ ╠═══════════════════════════════════════════════════════════════════════════╣ │ │ ║ ║ │ │ ║ Computes calibration metrics across multiple data segments ║ │ │ ║ ║ │ │ ║ ┌─────────────────────────────────────────────────────────────────────┐ ║ │ │ ║ │ Cached Properties │ ║ │ │ ║ ├─────────────────────────────────────────────────────────────────────┤ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_ecces_absolute ──────────▶ _ecce_per_segment() │ ║ │ │ ║ │ (ECCE per segment) │ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_sigmas ──────────────────▶ _ecce_standard_deviation_ │ ║ │ │ ║ │ (std dev per segment) per_segment() │ ║ │ │ ║ │ │ ║ │ │ ║ │ sigma_0 ─────────────────────────▶ _ecce_standard_deviation_ │ ║ │ │ ║ │ (global std dev) per_segment() │ ║ │ │ ║ │ │ ║ │ │ ║ │ p_value ─────────────────────────▶ ecce_pvalue_from_sigma() │ ║ │ │ ║ │ (min p-value across segments) │ ║ │ │ ║ │ │ ║ │ │ ║ │ segment_p_values ────────────────▶ ecce_pvalue_from_sigma() │ ║ │ │ ║ │ (p-value per segment) (vectorized) │ ║ │ │ ║ │ │ ║ │ │ ║ └─────────────────────────────────────────────────────────────────────┘ ║ │ │ ║ ║ │ │ ║ ┌─────────────────────────────────────────────────────────────────────┐ ║ │ │ ║ │ Derived Metrics │ ║ │ │ ║ ├─────────────────────────────────────────────────────────────────────┤ ║ │ │ ║ │ │ ║ │ │ ║ │ mce = max(segment_ecces_absolute) │ ║ │ │ ║ │ mce_sigma_scale = max(segment_ecces_absolute / segment_sigmas) │ ║ │ │ ║ │ mce_relative = mce / ecce_0 (normalized by global ECCE) │ ║ │ │ ║ │ │ ║ │ │ ║ └─────────────────────────────────────────────────────────────────────┘ ║ │ │ ║ ║ │ │ ╚═══════════════════════════════════════════════════════════════════════════╝ │ │ │ └─────────────────────────────────────────────────────────────────────────────────┘ ``` Reviewed By: Lorenzo-Perini Differential Revision: D91302777
…T_MCE (facebookincubator#169) Summary: Renames module-level constants, some variables and docstrings from legacy Kuiper naming to MCE (Multicalibration Error) naming for consistency with the ECCE API refactoring: - `DEFAULT_MULTI_KUIPER_MAX_VALUES_PER_SEGMENT_FEATURE` → `DEFAULT_MCE_MAX_VALUES_PER_SEGMENT_FEATURE` - `DEFAULT_MULTI_KUIPER_MIN_DEPTH` → `DEFAULT_MCE_MIN_DEPTH` - `DEFAULT_MULTI_KUIPER_MAX_DEPTH` → `DEFAULT_MCE_MAX_DEPTH` - `DEFAULT_MULTI_KUIPER_MIN_SAMPLES_PER_SEGMENT` → `DEFAULT_MCE_MIN_SAMPLES_PER_SEGMENT` - `DEFAULT_MULTI_KUIPER_GLOBAL_NORMALIZATION` → `DEFAULT_MCE_GLOBAL_NORMALIZATION` - `DEFAULT_MULTI_KUIPER_N_SEGMENTS` → `DEFAULT_MCE_N_SEGMENTS` Also updates fluent2 consumers to use the new constant names: - `fblearner/flow/projects/fluent2/definition/metrics/multicalibration_error.py` - `fblearner/flow/projects/fluent2/definition/evaluation.py` **Test Plan:** ``` buck2 test multicalibration/tests:test_metrics buck2 test fblearner/flow/projects/fluent2/workflows/evaluation/metrics/tests:multicalibration_error_test ``` Reviewed By: Lorenzo-Perini Differential Revision: D91306073
flinder
force-pushed
the
export-D91306073
branch
from
35f61df to
08eb111
Compare
flinder
added a commit
to flinder/MCGrad-1
that referenced
this pull request
Jan 26, 2026
…T_MCE (facebookincubator#169) Summary: Renames module-level constants, some variables and docstrings from legacy Kuiper naming to MCE (Multicalibration Error) naming for consistency with the ECCE API refactoring: - `DEFAULT_MULTI_KUIPER_MAX_VALUES_PER_SEGMENT_FEATURE` → `DEFAULT_MCE_MAX_VALUES_PER_SEGMENT_FEATURE` - `DEFAULT_MULTI_KUIPER_MIN_DEPTH` → `DEFAULT_MCE_MIN_DEPTH` - `DEFAULT_MULTI_KUIPER_MAX_DEPTH` → `DEFAULT_MCE_MAX_DEPTH` - `DEFAULT_MULTI_KUIPER_MIN_SAMPLES_PER_SEGMENT` → `DEFAULT_MCE_MIN_SAMPLES_PER_SEGMENT` - `DEFAULT_MULTI_KUIPER_GLOBAL_NORMALIZATION` → `DEFAULT_MCE_GLOBAL_NORMALIZATION` - `DEFAULT_MULTI_KUIPER_N_SEGMENTS` → `DEFAULT_MCE_N_SEGMENTS` Also updates fluent2 consumers to use the new constant names: - `fblearner/flow/projects/fluent2/definition/metrics/multicalibration_error.py` - `fblearner/flow/projects/fluent2/definition/evaluation.py` **Test Plan:** ``` buck2 test multicalibration/tests:test_metrics buck2 test fblearner/flow/projects/fluent2/workflows/evaluation/metrics/tests:multicalibration_error_test ``` Reviewed By: Lorenzo-Perini Differential Revision: D91306073
|
This pull request has been merged in d683f78. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
Summary:
Renames module-level constants, some variables and docstrings from legacy Kuiper naming to MCE (Multicalibration Error) naming for consistency with the ECCE API refactoring:
DEFAULT_MULTI_KUIPER_MAX_VALUES_PER_SEGMENT_FEATURE→DEFAULT_MCE_MAX_VALUES_PER_SEGMENT_FEATUREDEFAULT_MULTI_KUIPER_MIN_DEPTH→DEFAULT_MCE_MIN_DEPTHDEFAULT_MULTI_KUIPER_MAX_DEPTH→DEFAULT_MCE_MAX_DEPTHDEFAULT_MULTI_KUIPER_MIN_SAMPLES_PER_SEGMENT→DEFAULT_MCE_MIN_SAMPLES_PER_SEGMENTDEFAULT_MULTI_KUIPER_GLOBAL_NORMALIZATION→DEFAULT_MCE_GLOBAL_NORMALIZATIONDEFAULT_MULTI_KUIPER_N_SEGMENTS→DEFAULT_MCE_N_SEGMENTSAlso updates fluent2 consumers to use the new constant names:
fblearner/flow/projects/fluent2/definition/metrics/multicalibration_error.pyfblearner/flow/projects/fluent2/definition/evaluation.pyTest Plan:
Differential Revision: D91306073