Types¶

Type definitions and structured data models.

Type Definitions¶

TypedDict definitions for structured analysis results.

Type definitions for statqa package.

This module contains TypedDict definitions for structured data to provide better type safety than generic dict[str, Any].

class statqa.types.UnivariateResult[source]¶

Bases: TypedDict

Result of univariate analysis.

variable: str¶

label: str¶

variable_type: str¶

total_count: int¶

missing_count: int¶

missing_percentage: float¶

mean: float¶

median: float¶

std: float¶

min: float¶

max: float¶

q25: float¶

q75: float¶

iqr: float¶

skewness: float¶

kurtosis: float¶

robust_mean: float¶

mad: float¶

normality_test: dict[str, Any]¶

outliers: dict[str, Any]¶

mode: str | int¶

mode_count: int¶

unique_count: int¶

diversity_index: float¶

frequencies: dict[str, int]¶

analysis_type: Literal['numeric', 'categorical']¶

formatted_insight: str¶

class statqa.types.BivariateResult[source]¶

Bases: TypedDict

Result of bivariate analysis.

var1: str¶

var2: str¶

var1_label: str¶

var2_label: str¶

analysis_type: Literal['numeric_numeric', 'categorical_categorical', 'categorical_numeric']¶

sample_size: int¶

pearson: dict[str, Any]¶

spearman: dict[str, Any]¶

chi_square: dict[str, Any]¶

cramers_v: float¶

contingency_table: dict[str, Any]¶

t_test: dict[str, Any]¶

anova: dict[str, Any]¶

effect_size: float¶

effect_size_interpretation: str¶

significant: bool¶

formatted_insight: str¶

class statqa.types.QAPair[source]¶

Bases: TypedDict

Question-answer pair with provenance metadata.

question: str¶

answer: str¶

context: str¶

generated_at: str¶

tool: str¶

tool_version: str¶

generation_method: Literal['template', 'llm_paraphrase']¶

analysis_type: str¶

analyzer: str¶

llm_model: str | None¶

variable_name: str | None¶

variable_label: str | None¶

analysis_result: dict[str, Any]¶

class statqa.types.TemporalResult[source]¶

Bases: TypedDict

Result of temporal analysis.

variable: str¶

label: str¶

time_variable: str | None¶

analysis_type: Literal['temporal_trend', 'change_point', 'seasonality']¶

trend: Literal['increasing', 'decreasing', 'stable', 'insufficient_data']¶

tau: float¶

p_value: float¶

trend_significance: bool¶

change_points: list[dict[str, Any]]¶

seasonal_component: dict[str, Any]¶

formatted_insight: str¶

class statqa.types.CausalResult[source]¶

Bases: TypedDict

Result of causal analysis.

treatment: str¶

outcome: str¶

confounders: list[str]¶

analysis_type: Literal['treatment_effect', 'instrumental_variable', 'regression_discontinuity']¶

ate: float¶

ate_ci_lower: float¶

ate_ci_upper: float¶

effect_significant: bool¶

regression_results: dict[str, Any]¶

formatted_insight: str¶

class statqa.types.OpenAIFormat[source]¶

Bases: TypedDict

OpenAI fine-tuning format.

prompt: str¶

completion: str¶

class statqa.types.AnthropicFormat[source]¶

Bases: TypedDict

Anthropic fine-tuning format.

messages: list[dict[str, str]]¶

class statqa.types.JSONLFormat[source]¶

Bases: TypedDict

Standard JSONL format with provenance.

question: str¶

answer: str¶

metadata: dict[str, Any]¶