logml.survival_analysis.extractors.optimal_cut_off

Classes

OptimalCutOffSAExtractor(cfg, sa_setup, ...)

Runs optimal cuf-off analysis and builds plots on top of results.

class logml.survival_analysis.extractors.optimal_cut_off.OptimalCutOffSAParams

Bases: pydantic.main.BaseModel

Defines hyperparams for optimal cut-off method.

Show JSON schema

{
   "title": "OptimalCutOffSAParams",
   "description": "Defines hyperparams for optimal cut-off method.",
   "type": "object",
   "properties": {
      "n_plots_to_show": {
         "title": "N Plots To Show",
         "description": "Defines the number of Kaplan-Meier & Optimal Cut-Off plots that will be shown\n            in BaselineKit report.",
         "default": 10,
         "type": "integer"
      },
      "min_population": {
         "title": "Min Population",
         "description": "Defines the minimal fraction of samples that could be assigned to either low or high groups.",
         "default": 0.2,
         "type": "number"
      },
      "n_percentiles": {
         "title": "N Percentiles",
         "description": "Defines the number of thresholds within valid range that will be checked for \"optimality\"\n            in terms of Log Rank test statistic.",
         "default": 50,
         "type": "integer"
      }
   }
}

Fields

• min_population (float)

• n_percentiles (int)

• n_plots_to_show (int)

field n_plots_to_show: int = 10

Defines the number of Kaplan-Meier & Optimal Cut-Off plots that will be shown in BaselineKit report.

field min_population: float = 0.2

Defines the minimal fraction of samples that could be assigned to either low or high groups.

field n_percentiles: int = 50

Defines the number of thresholds within valid range that will be checked for “optimality” in terms of Log Rank test statistic.

class logml.survival_analysis.extractors.optimal_cut_off.OptimalCutOffSAExtractor(cfg: GlobalConfig, sa_setup: SurvivalAnalysisSetup, global_params: dict, logger=None)

Bases: logml.survival_analysis.extractors.base.BaseSAExtractor

Runs optimal cuf-off analysis and builds plots on top of results.

LABEL = 'optimal_cut_off'

CONFIG_CLASS

alias of logml.survival_analysis.extractors.optimal_cut_off.OptimalCutOffSAParams

run()

Two-step workflow:

• Runs optimal cut-off analysis to identify optimal thresholds for numerical grouping columns. Saves plots and summary.

• Runs KM analysis on top of the found thresholds.