nvflare.app_common.executors.statistics.statistics_task_handler module

class StatisticsTaskHandler(generator_id: str, precision: int = 4)

Bases: TaskHandler

StatisticsTaskHandler is to be used together with StatisticsExecutor.

StatisticsExecutor is client-side executor that perform local statistics generation and communication to FL Server global statistics controller. The actual local statistics calculation would delegate to Statistics spec implementor.

Init FLComponent.

The FLComponent is the base class of all FL Components. (executors, controllers, responders, filters, aggregators, and widgets are all FLComponents)

FLComponents have the capability to handle and fire events and contain various methods for logging.

execute_task(task_name: str, shareable: Shareable, fl_ctx: FLContext, abort_signal: Signal) → Shareable

Executes a task.

Parameters:

• task_name – task name

• shareable – input data

• fl_ctx – FLContext

• abort_signal (Signal) – signal to check during execution to determine whether this task is aborted.

Returns:

Output data

get_bin_range(feature_name: str, global_min_value: float, global_max_value: float, hist_config: dict) → List[float]

get_count(dataset_name: str, feature_name: str, statistic_configs: StatisticConfig, inputs: Shareable, fl_ctx: FLContext) → int

get_failure_count(dataset_name: str, feature_name: str, statistic_configs: StatisticConfig, inputs: Shareable, fl_ctx: FLContext) → int

get_histogram(dataset_name: str, feature_name: str, statistic_configs: StatisticConfig, inputs: Shareable, fl_ctx: FLContext) → Histogram

get_max_value(dataset_name: str, feature_name: str, statistic_configs: StatisticConfig, inputs: Shareable, fl_ctx: FLContext) → float

get randomized max value

get_mean(dataset_name: str, feature_name: str, statistic_configs: StatisticConfig, inputs: Shareable, fl_ctx: FLContext) → float

get_min_value(dataset_name: str, feature_name: str, statistic_configs: StatisticConfig, inputs: Shareable, fl_ctx: FLContext) → float

get randomized min value

get_number_of_bins(feature_name: str, hist_config: dict) → int

get_numeric_features() → Dict[str, List[Feature]]

get_stddev(dataset_name: str, feature_name: str, statistic_configs: StatisticConfig, inputs: Shareable, fl_ctx: FLContext) → float

get_sum(dataset_name: str, feature_name: str, statistic_configs: StatisticConfig, inputs: Shareable, fl_ctx: FLContext) → float

get_variance_with_mean(dataset_name: str, feature_name: str, statistic_configs: StatisticConfig, inputs: Shareable, fl_ctx: FLContext) → float

initialize(fl_ctx: FLContext)

This is called when client is start Run. At this point the server hasn’t communicated to the local component yet. :param fl_ctx: fl_ctx: FLContext of the running environment

pre_run(target_statistics: List[StatisticConfig])

statistic_functions() → dict