diff --git a/pyproject.toml b/pyproject.toml index 8e22790..9ccca90 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -4,7 +4,7 @@ build-backend = "hatchling.build" [project] name = "scorepilot" -version = "0.22.0" +version = "0.22.1" description = "Web-based tool for PCA/PLS model analysis in chemometrics." readme = "README.md" requires-python = ">=3.12" diff --git a/src/scorepilot/core/profiling.py b/src/scorepilot/core/profiling.py index 74f0540..17d5974 100644 --- a/src/scorepilot/core/profiling.py +++ b/src/scorepilot/core/profiling.py @@ -142,9 +142,11 @@ def sequence(series: pd.Series) -> list[float | None]: def suggest_transform(summary: VariableSummary) -> TransformKind: """Suggest a transform from a variable's shape. - A strongly right-skewed, strictly positive variable with a wide dynamic range - is a candidate for a log transform; a milder skew suggests a signed power - (root). Otherwise no transform is suggested. + A strongly right-skewed (skewness > 1.0), strictly positive variable with a + wide dynamic range (min/max ratio > 20) is a candidate for a log transform. + A strongly right-skewed variable that is not strictly positive or lacks a + wide dynamic range falls back to a signed power (root). Otherwise no + transform is suggested. """ if summary.column_type is not ColumnType.QUANTITATIVE or summary.skewness is None: return TransformKind.NONE