hossam 0.4.3__py3-none-any.whl → 0.4.5__py3-none-any.whl

hossam/hs_stats.py

@@ -1,5 +1,7 @@
 # -*- coding: utf-8 -*-
+
 from __future__ import annotations
+from typing import overload, Tuple, Literal, Union, Any
 
 # -------------------------------------------------------------
 import numpy as np
@@ -22,12 +24,13 @@ from scipy.stats import (
     normaltest,
     bartlett,
     levene,
-    ttest_1samp,
+    ttest_1samp,    # type: ignore
     ttest_ind as scipy_ttest_ind,
-    ttest_rel,
+    ttest_rel,      # type: ignore
     wilcoxon,
     pearsonr,
     spearmanr,
+    chi2
 )
 
 import statsmodels.api as sm
@@ -36,10 +39,71 @@ from statsmodels.stats.outliers_influence import variance_inflation_factor
 from statsmodels.stats.multitest import multipletests
 from statsmodels.stats.stattools import durbin_watson
 from statsmodels.regression.linear_model import RegressionResultsWrapper
+from statsmodels.discrete.discrete_model import BinaryResultsWrapper
 from statsmodels.discrete.discrete_model import BinaryResults
 
 from pingouin import anova, pairwise_tukey, welch_anova, pairwise_gameshowell
 
+from .hs_plot import ols_residplot, ols_qqplot
+
+# ===================================================================
+# MCAR(결측치 무작위성) 검정
+# ===================================================================
+def mcar_test(data: DataFrame, columns: list | str | None = None) -> DataFrame:
+    if isinstance(columns, str):
+        columns = [c.strip() for c in columns.split(",")]
+
+    cols = data.columns if columns is None else columns
+    df = data[cols]
+
+    # 결측치가 있는 컬럼만 사용
+    cols_with_na = [c for c in df.columns if df[c].isna().any()]
+    if len(cols_with_na) < 2:
+        raise ValueError("MCAR 검정은 결측치가 있는 변수가 최소 2개 이상 필요합니다.")
+
+    X = df[cols_with_na].to_numpy()
+    n, p = X.shape
+
+    # complete cases로 평균·공분산 추정
+    complete = ~np.isnan(X).any(axis=1)
+    if complete.sum() < p + 1:
+        raise ValueError("완전관측치(complete cases)가 부족하여 MCAR 검정을 수행할 수 없습니다.")
+
+    mu = X[complete].mean(axis=0)
+    S = np.cov(X[complete], rowvar=False)
+    S_inv = np.linalg.pinv(S)
+
+    chi_sq = 0.0
+    dfree = 0
+
+    for i in range(n):
+        obs = ~np.isnan(X[i])
+        if obs.sum() == p:
+            continue  # complete case는 제외
+        diff = X[i, obs] - mu[obs]
+        S_obs = S[np.ix_(obs, obs)]
+        S_obs_inv = np.linalg.pinv(S_obs)
+
+        chi_sq += diff @ S_obs_inv @ diff
+        dfree += obs.sum()
+
+    dfree -= p  # Little's adjustment
+
+    p_value = 1 - chi2.cdf(chi_sq, dfree)
+    is_mcar = p_value > 0.05
+
+    return DataFrame([{
+        "statistic": chi_sq,
+        "dof": dfree,
+        "p-value": p_value,
+        "is_mcar": is_mcar,
+        "interpretation": (
+            "결측치는 완전 무작위(MCAR)로 판단됨 → 결측치 삭제 가능"
+            if is_mcar else
+            "결측치는 완전 무작위(MCAR)가 아님 → 삭제 시 편향 가능"
+        )
+    }])
+
 # ===================================================================
 # 결측치 분석 (Missing Values Analysis)
 # ===================================================================
@@ -219,6 +283,8 @@ def describe(data: DataFrame, *fields: str, columns: list | None = None):
             행은 다음과 같은 통계량을 포함:
 
             - count (float): 비결측치의 수
+            - na_count (int): 결측치의 수
+            - na_rate (float): 결측치 비율(%)
             - mean (float): 평균값
             - std (float): 표준편차
             - min (float): 최소값
@@ -267,9 +333,13 @@ def describe(data: DataFrame, *fields: str, columns: list | None = None):
 
     # 기술통계량 구하기
     desc = data[list(fields)].describe().T
-    # 각 컬럼별 결측치 수(null_count) 추가
-    null_counts = data[list(fields)].isnull().sum()
-    desc.insert(1, 'null_count', null_counts)
+
+    # 각 컬럼별 결측치 수(na_count) 추가
+    na_counts = data[list(fields)].isnull().sum()
+    desc.insert(1, 'na_count', na_counts)
+
+    # 결측치 비율(na_rate) 추가
+    desc.insert(2, 'na_rate', (na_counts / len(data)) * 100)
 
     # 추가 통계량 계산
     additional_stats = []
@@ -305,29 +375,29 @@ def describe(data: DataFrame, *fields: str, columns: list | None = None):
         outlier_rate = (outlier_count / len(data)) * 100
 
         # 분포 특성 판정 (왜도 기준)
-        abs_skew = abs(skew)
-        if abs_skew < 0.5:
+        abs_skew = abs(skew)    # type: ignore
+        if abs_skew < 0.5:      # type: ignore
             dist = "거의 대칭"
-        elif abs_skew < 1.0:
-            if skew > 0:
+        elif abs_skew < 1.0:    # type: ignore
+            if skew > 0:        # type: ignore
                 dist = "약한 우측 꼬리"
             else:
                 dist = "약한 좌측 꼬리"
-        elif abs_skew < 2.0:
-            if skew > 0:
+        elif abs_skew < 2.0:    # type: ignore
+            if skew > 0:        # type: ignore
                 dist = "중간 우측 꼬리"
             else:
                 dist = "중간 좌측 꼬리"
         else:
-            if skew > 0:
+            if skew > 0:        # type: ignore
                 dist = "극단 우측 꼬리"
             else:
                 dist = "극단 좌측 꼬리"
 
         # 로그변환 필요성 판정
-        if abs_skew < 0.5:
+        if abs_skew < 0.5:      # type: ignore
             log_need = "낮음"
-        elif abs_skew < 1.0:
+        elif abs_skew < 1.0:    # type: ignore
             log_need = "중간"
         else:
             log_need = "높음"
@@ -403,7 +473,7 @@ def category_describe(data: DataFrame, *fields: str):
     """
     if not fields:
         # 명목형(범주형) 컬럼 선택: object, category, bool 타입
-        fields = data.select_dtypes(include=['object', 'category', 'bool']).columns
+        fields = data.select_dtypes(include=['object', 'category', 'bool']).columns # type: ignore
 
     result = []
     summary = []
@@ -660,7 +730,7 @@ def equal_var_test(data: DataFrame, columns: list | str | None = None, normal_di
         normality_result = normal_test(data[numeric_cols], method="n")
         # 모든 컬럼이 정규분포를 따르는지 확인
         all_normal = normality_result["is_normal"].all()
-        normal_dist = all_normal
+        normal_dist = all_normal    # type: ignore
 
     try:
         if normal_dist:
@@ -759,7 +829,7 @@ def ttest_1samp(data, mean_value: float = 0.0) -> DataFrame:
     else:
         for a in alternative:
             try:
-                s, p = ttest_1samp(col_data, mean_value, alternative=a)
+                s, p = ttest_1samp(col_data, mean_value, alternative=a) # type: ignore
 
                 itp = None
 
@@ -869,26 +939,26 @@ def ttest_ind(x, y, equal_var: bool | None = None) -> DataFrame:
 
     for a in alternative:
         try:
-            s, p = scipy_ttest_ind(x_data, y_data, equal_var=equal_var, alternative=a)
+            s, p = scipy_ttest_ind(x_data, y_data, equal_var=equal_var, alternative=a)  # type: ignore
             n = "t-test_ind" if equal_var else "Welch's t-test"
 
             # 검정 결과 해석
             itp = None
 
             if a == "two-sided":
-                itp = fmt.format("==" if p > 0.05 else "!=")
+                itp = fmt.format("==" if p > 0.05 else "!=")    # type: ignore
             elif a == "less":
-                itp = fmt.format(">=" if p > 0.05 else "<")
+                itp = fmt.format(">=" if p > 0.05 else "<")     # type: ignore
             else:
-                itp = fmt.format("<=" if p > 0.05 else ">")
+                itp = fmt.format("<=" if p > 0.05 else ">")     # type: ignore
 
             result.append({
                 "test": n,
                 "alternative": a,
-                "statistic": round(s, 3),
-                "p-value": round(p, 4),
-                "H0": p > 0.05,
-                "H1": p <= 0.05,
+                "statistic": round(s, 3),   # type: ignore
+                "p-value": round(p, 4),     # type: ignore
+                "H0": p > 0.05,             # type: ignore
+                "H1": p <= 0.05,            # type: ignore
                 "interpretation": itp,
                 "equal_var_checked": var_checked
             })
@@ -998,7 +1068,7 @@ def ttest_rel(x, y, parametric: bool | None = None) -> DataFrame:
     for a in alternative:
         try:
             if parametric:
-                s, p = ttest_rel(x_data, y_data, alternative=a)
+                s, p = ttest_rel(x_data, y_data, alternative=a) # type: ignore
                 n = "t-test_paired"
             else:
                 # Wilcoxon signed-rank test (대응표본용 비모수 검정)
@@ -1008,19 +1078,19 @@ def ttest_rel(x, y, parametric: bool | None = None) -> DataFrame:
             itp = None
 
             if a == "two-sided":
-                itp = fmt.format("==" if p > 0.05 else "!=")
+                itp = fmt.format("==" if p > 0.05 else "!=")    # type: ignore
             elif a == "less":
-                itp = fmt.format(">=" if p > 0.05 else "<")
+                itp = fmt.format(">=" if p > 0.05 else "<")     # type: ignore
             else:
-                itp = fmt.format("<=" if p > 0.05 else ">")
+                itp = fmt.format("<=" if p > 0.05 else ">")     # type: ignore
 
             result.append({
                 "test": n,
                 "alternative": a,
-                "statistic": round(s, 3) if not np.isnan(s) else s,
-                "p-value": round(p, 4) if not np.isnan(p) else p,
-                "H0": p > 0.05,
-                "H1": p <= 0.05,
+                "statistic": round(s, 3) if not np.isnan(s) else s, # type: ignore
+                "p-value": round(p, 4) if not np.isnan(p) else p,   # type: ignore
+                "H0": p > 0.05,     # type: ignore
+                "H1": p <= 0.05,    # type: ignore
                 "interpretation": itp,
                 "normality_checked": var_checked
             })
@@ -1047,7 +1117,7 @@ def ttest_rel(x, y, parametric: bool | None = None) -> DataFrame:
 # ===================================================================
 def vif_filter(
     data: DataFrame,
-    yname: str = None,
+    yname: str | None = None,
     ignore: list | None = None,
     threshold: float = 10.0,
     verbose: bool = False,
@@ -1112,7 +1182,7 @@ def vif_filter(
         for i, col in enumerate(X_clean.columns, start=0):
             # exog의 첫 열은 상수항이므로 변수 인덱스는 +1
             try:
-                vifs[col] = float(variance_inflation_factor(exog.values, i + 1))
+                vifs[col] = float(variance_inflation_factor(exog.values, i + 1))# type: ignore
             except Exception:
                 # 계산 실패 시 무한대로 처리하여 우선 제거 대상으로
                 vifs[col] = float("inf")
@@ -1150,7 +1220,7 @@ def vif_filter(
 # ===================================================================
 # x, y 데이터에 대한 추세선을 구한다.
 # ===================================================================
-def trend(x: any, y: any, degree: int = 1, value_count: int = 100) -> Tuple[np.ndarray, np.ndarray]:
+def trend(x: Any, y: Any, degree: int = 1, value_count: int = 100) -> Tuple[np.ndarray, np.ndarray]:
     """x, y 데이터에 대한 추세선을 구한다.
 
     Args:
@@ -1192,7 +1262,10 @@ def trend(x: any, y: any, degree: int = 1, value_count: int = 100) -> Tuple[np.n
 # ===================================================================
 # 선형회귀 요약 리포트
 # ===================================================================
-def ols_report(fit, data, full=False, alpha=0.05):
+def ols_report(fit, data, full=False, alpha=0.05) -> Union[
+    Tuple[DataFrame, DataFrame],
+    Tuple[DataFrame, DataFrame, str, str, list[str], str]
+]:
     """선형회귀 적합 결과를 요약 리포트로 변환한다.
 
     Args:
@@ -1211,6 +1284,7 @@ def ols_report(fit, data, full=False, alpha=0.05):
             - 회귀식 문자열 (`equation_text`, str): 상수항과 계수를 포함한 회귀식 표현.
 
         full=False일 때:
+            - 성능 지표 표 (`pdf`, DataFrame): R, R², Adj. R², F, p-value, Durbin-Watson.
             - 회귀계수 표 (`rdf`, DataFrame)
 
     Examples:
@@ -1250,7 +1324,7 @@ def ols_report(fit, data, full=False, alpha=0.05):
     for i, col in enumerate(indi_df.columns, start=1):  # 상수항이 0이므로 1부터 시작
         try:
             with np.errstate(divide='ignore', invalid='ignore'):
-                vif_value = variance_inflation_factor(indi_df_const.values, i)
+                vif_value = variance_inflation_factor(indi_df_const.values, i)  # type: ignore
                 # inf나 매우 큰 값 처리
                 if np.isinf(vif_value) or vif_value > 1e10:
                     vif_dict[col] = np.inf
@@ -1378,7 +1452,19 @@ def ols_report(fit, data, full=False, alpha=0.05):
 # ===================================================================
 # 선형회귀
 # ===================================================================
-def ols(df: DataFrame, yname: str, report=False):
+def ols(df: DataFrame, yname: str, report: bool | str | int = False) -> Union[
+    RegressionResultsWrapper,
+    Tuple[RegressionResultsWrapper, DataFrame, DataFrame],
+    Tuple[
+        RegressionResultsWrapper,
+        DataFrame,
+        DataFrame,
+        str,
+        str,
+        list[str],
+        str
+    ]
+]:
     """선형회귀분석을 수행하고 적합 결과를 반환한다.
 
     OLS(Ordinary Least Squares) 선형회귀분석을 실시한다.
@@ -1387,7 +1473,7 @@ def ols(df: DataFrame, yname: str, report=False):
     Args:
         df (DataFrame): 종속변수와 독립변수를 모두 포함한 데이터프레임.
         yname (str): 종속변수 컬럼명.
-        report: 리포트 모드 설정. 다음 값 중 하나:
+        report (bool | str | int): 리포트 모드 설정. 다음 값 중 하나:
             - False (기본값): 리포트 미사용. fit 객체만 반환.
             - 1 또는 'summary': 요약 리포트 반환 (full=False).
             - 2 또는 'full': 풀 리포트 반환 (full=True).
@@ -1426,10 +1512,10 @@ def ols(df: DataFrame, yname: str, report=False):
         fit = hs_stats.ols(df, 'target')
 
         # 요약 리포트 반환
-        fit, pdf, rdf = hs_stats.ols(df, 'target', report=1)
+        fit, pdf, rdf = hs_stats.ols(df, 'target', report='summary')
 
         # 풀 리포트 반환
-        fit, pdf, rdf, result_report, model_report, var_reports, eq = hs_stats.ols(df, 'target', report=2)
+        fit, pdf, rdf, result_report, model_report, var_reports, eq = hs_stats.ols(df, 'target', report='full')
         ```
     """
     x = df.drop(yname, axis=1)
@@ -1445,11 +1531,11 @@ def ols(df: DataFrame, yname: str, report=False):
         return linear_fit
     elif report == 1 or report == 'summary':
         # 요약 리포트 (full=False)
-        pdf, rdf = ols_report(linear_fit, df, full=False, alpha=0.05)
+        pdf, rdf = ols_report(linear_fit, df, full=False, alpha=0.05)   # type: ignore
         return linear_fit, pdf, rdf
     elif report == 2 or report == 'full' or report is True:
         # 풀 리포트 (full=True)
-        pdf, rdf, result_report, model_report, variable_reports, equation_text = ols_report(linear_fit, df, full=True, alpha=0.05)
+        pdf, rdf, result_report, model_report, variable_reports, equation_text = ols_report(linear_fit, df, full=True, alpha=0.05)  # type: ignore
         return linear_fit, pdf, rdf, result_report, model_report, variable_reports, equation_text
     else:
         # 기본값: 리포트 미사용
@@ -1459,15 +1545,31 @@ def ols(df: DataFrame, yname: str, report=False):
 # ===================================================================
 # 로지스틱 회귀 요약 리포트
 # ===================================================================
-def logit_report(fit, data, threshold=0.5, full=False, alpha=0.05):
+def logit_report(
+    fit: BinaryResultsWrapper,
+    data: DataFrame,
+    threshold: float = 0.5,
+    full: Union[bool, str, int] = False,
+    alpha: float = 0.05
+) -> Union[
+    Tuple[DataFrame, DataFrame],
+    Tuple[
+        DataFrame,
+        DataFrame,
+        str,
+        str,
+        list[str],
+        np.ndarray
+    ]
+]:
     """로지스틱 회귀 적합 결과를 상세 리포트로 변환한다.
 
     Args:
         fit: statsmodels Logit 결과 객체 (`fit.summary()`와 예측 확률을 지원해야 함).
-        data: 종속변수와 독립변수를 모두 포함한 DataFrame.
-        threshold: 예측 확률을 이진 분류로 변환할 임계값. 기본값 0.5.
-        full: True이면 6개 값 반환, False이면 주요 2개(cdf, rdf)만 반환. 기본값 False.
-        alpha: 유의수준. 기본값 0.05.
+        data (DataFrame): 종속변수와 독립변수를 모두 포함한 DataFrame.
+        threshold (float): 예측 확률을 이진 분류로 변환할 임계값. 기본값 0.5.
+        full (bool | str | int): True이면 6개 값 반환, False이면 주요 2개(cdf, rdf)만 반환. 기본값 False.
+        alpha (float): 유의수준. 기본값 0.05.
 
     Returns:
         tuple: full=True일 때 다음 요소를 포함한다.
@@ -1555,7 +1657,7 @@ def logit_report(fit, data, threshold=0.5, full=False, alpha=0.05):
     vif_dict = {}
     x_const = sm.add_constant(x, has_constant="add")
     for i, col in enumerate(x.columns, start=1):  # 상수항이 0이므로 1부터 시작
-        vif_dict[col] = variance_inflation_factor(x_const.values, i)
+        vif_dict[col] = variance_inflation_factor(x_const.values, i)    # type: ignore
 
     for idx, row in tbl.iterrows():
         name = idx
@@ -1652,7 +1754,25 @@ def logit_report(fit, data, threshold=0.5, full=False, alpha=0.05):
 # ===================================================================
 # 로지스틱 회귀
 # ===================================================================
-def logit(df: DataFrame, yname: str, report=False):
+def logit(
+    df: DataFrame,
+    yname: str,
+    report: Union[bool, str, int] = False
+) -> Union[
+    BinaryResultsWrapper,
+    Tuple[
+        BinaryResultsWrapper,
+        DataFrame
+    ],
+    Tuple[
+        BinaryResultsWrapper,
+        DataFrame,
+        DataFrame,
+        str,
+        str,
+        list[str]
+    ]
+]:
     """로지스틱 회귀분석을 수행하고 적합 결과를 반환한다.
 
     종속변수가 이항(binary) 형태일 때 로지스틱 회귀분석을 실시한다.
@@ -1718,13 +1838,13 @@ def logit(df: DataFrame, yname: str, report=False):
         return logit_fit
     elif report == 1 or report == 'summary':
         # 요약 리포트 (full=False)
-        cdf, rdf = logit_report(logit_fit, df, threshold=0.5, full=False, alpha=0.05)
+        cdf, rdf = logit_report(logit_fit, df, threshold=0.5, full=False, alpha=0.05)   # type: ignore
         # 요약에서는 result_report와 variable_reports만 포함
         # 간단한 버전으로 result와 variable_reports만 생성
         return logit_fit, rdf
     elif report == 2 or report == 'full' or report is True:
         # 풀 리포트 (full=True)
-        cdf, rdf, result_report, model_report, variable_reports, cm = logit_report(logit_fit, df, threshold=0.5, full=True, alpha=0.05)
+        cdf, rdf, result_report, model_report, variable_reports, cm = logit_report(logit_fit, df, threshold=0.5, full=True, alpha=0.05) # type: ignore
         return logit_fit, cdf, rdf, result_report, model_report, variable_reports
     else:
         # 기본값: 리포트 미사용
@@ -1734,7 +1854,7 @@ def logit(df: DataFrame, yname: str, report=False):
 # ===================================================================
 # 선형성 검정 (Linearity Test)
 # ===================================================================
-def ols_linearity_test(fit, power: int = 2, alpha: float = 0.05) -> DataFrame:
+def ols_linearity_test(fit, power: int = 2, alpha: float = 0.05, plot: bool = False, title: str | None = None, save_path: str | None = None) -> DataFrame:
     """회귀모형의 선형성을 Ramsey RESET 검정으로 평가한다.
 
     적합된 회귀모형에 대해 Ramsey RESET(Regression Specification Error Test) 검정을 수행하여
@@ -1747,6 +1867,9 @@ def ols_linearity_test(fit, power: int = 2, alpha: float = 0.05) -> DataFrame:
                                power=2일 때 예측값의 제곱항이 추가됨.
                                power가 클수록 더 높은 차수의 비선형성을 감지.
         alpha (float, optional): 유의수준. 기본값 0.05.
+        plot (bool, optional): True이면 잔차 플롯을 출력. 기본값 False.
+        title (str, optional): 플롯 제목. 기본값 None.
+        save_path (str, optional): 플롯을 저장할 경로. 기본값 None
 
     Returns:
         DataFrame: 선형성 검정 결과를 포함한 데이터프레임.
@@ -1829,13 +1952,16 @@ def ols_linearity_test(fit, power: int = 2, alpha: float = 0.05) -> DataFrame:
         "해석": [interpretation]
     })
 
+    if plot:
+        ols_residplot(fit, lowess=True, mse=True, title=title, save_path=save_path)
+
     return result_df
 
 
 # ===================================================================
 # 정규성 검정 (Normality Test)
 # ===================================================================
-def ols_normality_test(fit, alpha: float = 0.05) -> DataFrame:
+def ols_normality_test(fit, alpha: float = 0.05, plot: bool = False, title: str | None = None, save_path: str | None = None) -> DataFrame:
     """회귀모형 잔차의 정규성을 검정한다.
 
     회귀모형의 잔차가 정규분포를 따르는지 Shapiro-Wilk 검정과 Jarque-Bera 검정으로 평가한다.
@@ -1844,6 +1970,9 @@ def ols_normality_test(fit, alpha: float = 0.05) -> DataFrame:
     Args:
         fit: 회귀 모형 객체 (statsmodels의 RegressionResultsWrapper).
         alpha (float, optional): 유의수준. 기본값 0.05.
+        plot (bool, optional): True이면 Q-Q 플롯을 출력. 기본값 False.
+        title (str, optional): 플롯 제목. 기본값 None.
+        save_path (str, optional): 플롯을 저장할 경로. 기본값 None
 
     Returns:
         DataFrame: 정규성 검정 결과를 포함한 데이터프레임.
@@ -1900,7 +2029,7 @@ def ols_normality_test(fit, alpha: float = 0.05) -> DataFrame:
     # 2. Jarque-Bera 검정 (항상 수행)
     try:
         stat_jb, p_jb = jarque_bera(residuals)
-        significant_jb = p_jb <= alpha
+        significant_jb = p_jb <= alpha  # type: ignore
 
         if significant_jb:
             interpretation_jb = f"정규성 위반 (p={p_jb:.4f} <= {alpha})"
@@ -1922,6 +2051,10 @@ def ols_normality_test(fit, alpha: float = 0.05) -> DataFrame:
     if not results:
         raise ValueError("정규성 검정을 수행할 수 없습니다.")
 
+
+    if plot:
+        ols_qqplot(fit, title=title, save_path=save_path)
+
     result_df = DataFrame(results)
     return result_df
 
@@ -2229,8 +2362,8 @@ def corr_pairwise(
             corr_val, pval = np.nan, np.nan
 
         # 4) 유의성, 강도
-        significant = False if np.isnan(pval) else pval <= alpha
-        abs_r = abs(corr_val) if not np.isnan(corr_val) else 0
+        significant = False if np.isnan(pval) else pval <= alpha    # type: ignore
+        abs_r = abs(corr_val) if not np.isnan(corr_val) else 0      # type: ignore
         if abs_r > 0.7:
             strength = "strong"
         elif abs_r > 0.3:
@@ -2397,13 +2530,13 @@ def oneway_anova(data: DataFrame, dv: str, between: str, alpha: float = 0.05) ->
         anova_df['significant'] = anova_df['p-unc'] <= alpha
 
     # ANOVA 결과가 유의한지 확인
-    p_unc = float(anova_df.loc[0, 'p-unc'])
+    p_unc = float(anova_df.loc[0, 'p-unc']) # type: ignore
     anova_significant = p_unc <= alpha
 
     # ANOVA 보고 문장 생성
     def _safe_get(col: str, default: float = np.nan) -> float:
         try:
-            return float(anova_df.loc[0, col]) if col in anova_df.columns else default
+            return float(anova_df.loc[0, col]) if col in anova_df.columns else default  # type: ignore
         except Exception:
             return default
 
@@ -2718,7 +2851,7 @@ def predict(fit, data: DataFrame | Series) -> DataFrame | Series | float:
 
         # Series 입력인 경우 단일 값 반환
         if is_series:
-            return float(predictions.iloc[0])
+            return float(predictions.iloc[0])   # type: ignore
 
         # DataFrame 입력인 경우
         if isinstance(data, DataFrame):
@@ -2791,8 +2924,7 @@ def corr_effect_size(data: DataFrame, dv: str, *fields: str, alpha: float = 0.05
 
     # fields가 지정되지 않으면 수치형 컬럼 중 dv 제외 모두 사용
     if not fields:
-        fields = [col for col in data.columns
-                 if is_numeric_dtype(data[col]) and col != dv]
+        fields = [col for col in data.columns if is_numeric_dtype(data[col]) and col != dv] # type: ignore
 
     # dv가 수치형인지 확인
     if not is_numeric_dtype(data[dv]):
@@ -2820,8 +2952,8 @@ def corr_effect_size(data: DataFrame, dv: str, *fields: str, alpha: float = 0.05
         normal_y_result = normal_test(data[[dv]], columns=[dv], method=method_y)
 
         # 정규성 판정 (p > alpha면 정규분포 가정)
-        normal_x = normal_x_result.loc[var, 'p-val'] > alpha if var in normal_x_result.index else False
-        normal_y = normal_y_result.loc[dv, 'p-val'] > alpha if dv in normal_y_result.index else False
+        normal_x = normal_x_result.loc[var, 'p-val'] > alpha if var in normal_x_result.index else False     # type: ignore
+        normal_y = normal_y_result.loc[dv, 'p-val'] > alpha if dv in normal_y_result.index else False   # type: ignore
 
         # Pearson (모두 정규) vs Spearman (하나라도 비정규)
         if normal_x and normal_y:
@@ -2833,8 +2965,8 @@ def corr_effect_size(data: DataFrame, dv: str, *fields: str, alpha: float = 0.05
 
         # Cohen's d 계산 (상관계수에서 효과크기로 변환)
         # d = 2*r / sqrt(1-r^2)
-        if r**2 < 1:
-            d = (2 * r) / np.sqrt(1 - r**2)
+        if r ** 2 < 1:    # type: ignore
+            d = (2 * r) / np.sqrt(1 - r ** 2) # type: ignore
         else:
             d = 0
 

hossam/hs_util.py

@@ -122,11 +122,9 @@ def pretty_table(data: DataFrame, tablefmt="simple", headers: str = "keys") -> N
         ```
     """
 
-    tabulate.WIDE_CHARS_MODE = False
+    tabulate.WIDE_CHARS_MODE = False # type: ignore
     print(
-        tabulate(
-            data, headers=headers, tablefmt=tablefmt, showindex=True, numalign="right"
-        )
+        tabulate(data, headers=headers, tablefmt=tablefmt, showindex=True, numalign="right") # type: ignore
     )
 
 
@@ -167,7 +165,7 @@ def __data_info(
 
     if info:
         print("\n✅ 테이블 정보")
-        pretty_table(data.info(), tablefmt="pretty")
+        pretty_table(data.info(), tablefmt="pretty") # type: ignore
 
         print("\n✅ 상위 5개 행")
         pretty_table(data.head(), tablefmt="pretty")
@@ -229,7 +227,7 @@ def load_data(key: str,
     elif k.endswith(".csv"):
         origin = read_csv(key)
     else:
-        origin = _load_data_remote(key, local)
+        origin = _load_data_remote(key, local) # type: ignore
 
     if origin is None:
         raise RuntimeError("Data loading failed: origin is None")

{hossam-0.4.3.dist-info → hossam-0.4.5.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: hossam
-Version: 0.4.3
+Version: 0.4.5
 Summary: Hossam Data Helper
 Author-email: Lee Kwang-Ho <leekh4232@gmail.com>
 License-Expression: MIT

hossam-0.4.5.dist-info/RECORD

@@ -0,0 +1,16 @@
+hossam/NotoSansKR-Regular.ttf,sha256=0SCufUQwcVWrWTu75j4Lt_V2bgBJIBXl1p8iAJJYkVY,6185516
+hossam/__init__.py,sha256=OkMeP15jt6aCy7QNXMtkO0YRVvgOQYumkb7GuVKrbcs,2712
+hossam/data_loader.py,sha256=K0-MJaVeedF5x8mSp22X2rD_CZ-T185EhoUFEqzP8Ss,6352
+hossam/hs_classroom.py,sha256=rgayol3U5PSo4rLfdbClfiAtG21bFrASaSW56PUsjus,27144
+hossam/hs_gis.py,sha256=DVmndBK-_7GMK3J1_on3ieEQk1S0MfUZ8_wlX-cDdZQ,11581
+hossam/hs_plot.py,sha256=3j9B69pl-zQM_09lTXxLKAMaDM0vwOTsUWbzcU8hCK8,86228
+hossam/hs_prep.py,sha256=kCmFxnMyFZ5tLUfoE8msbwTracajHAmruJbFj6A6eIU,38020
+hossam/hs_stats.py,sha256=uGYkEk8Rb8qMoZ5FiZ7Yg6jssLIGl_EBbmwvvSYljhQ,115780
+hossam/hs_timeserise.py,sha256=gSj3cPgOGLOZEXhfW1anXbwpoJja847ZY9F8l9piJPE,42601
+hossam/hs_util.py,sha256=xuNXC6FJSAmyAbcRAUMsigCKHXM25t3H90nFMgq7IBs,8482
+hossam/leekh.png,sha256=1PB5NQ24SDoHA5KMiBBsWpSa3iniFcwFTuGwuOsTHfI,6395
+hossam-0.4.5.dist-info/licenses/LICENSE,sha256=nIqzhlcFY_2D6QtFsYjwU7BWkafo-rUJOQpDZ-DsauI,941
+hossam-0.4.5.dist-info/METADATA,sha256=HM5qrrvaFZWAyUlhgV_BLPHAcxEZdZ4gp2p3V4X4pzo,3676
+hossam-0.4.5.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+hossam-0.4.5.dist-info/top_level.txt,sha256=_-7bwjhthHplWhywEaHIJX2yL11CQCaLjCNSBlk6wiQ,7
+hossam-0.4.5.dist-info/RECORD,,

hossam-0.4.3.dist-info/RECORD

@@ -1,16 +0,0 @@
-hossam/NotoSansKR-Regular.ttf,sha256=0SCufUQwcVWrWTu75j4Lt_V2bgBJIBXl1p8iAJJYkVY,6185516
-hossam/__init__.py,sha256=OkMeP15jt6aCy7QNXMtkO0YRVvgOQYumkb7GuVKrbcs,2712
-hossam/data_loader.py,sha256=oUIsqbHQoRiHA_1tdElDaYo1ipmUB5fYSXYMB5gLOl0,6395
-hossam/hs_classroom.py,sha256=rgayol3U5PSo4rLfdbClfiAtG21bFrASaSW56PUsjus,27144
-hossam/hs_gis.py,sha256=DLogaf5nxJBbG-d8QoH2g8UfZ1omMtmEXDYgNg8jtT0,11410
-hossam/hs_plot.py,sha256=A_nS8dP4cijp7LZs253SWxfBUp5qvvTlSPGKjDj0BIA,83712
-hossam/hs_prep.py,sha256=2ptFFxV4G1IFmy-B89TqXaPkA8jROZutr2XIkaXNHW4,36006
-hossam/hs_stats.py,sha256=qAor-RE5qNsytoZW1mriK3yql9PVif5bBGyG64YC2PM,110780
-hossam/hs_timeserise.py,sha256=gSj3cPgOGLOZEXhfW1anXbwpoJja847ZY9F8l9piJPE,42601
-hossam/hs_util.py,sha256=8byLj_VR93vS__lyf0xgQKArgMy9qFm2VvZVSCxfQX0,8444
-hossam/leekh.png,sha256=1PB5NQ24SDoHA5KMiBBsWpSa3iniFcwFTuGwuOsTHfI,6395
-hossam-0.4.3.dist-info/licenses/LICENSE,sha256=nIqzhlcFY_2D6QtFsYjwU7BWkafo-rUJOQpDZ-DsauI,941
-hossam-0.4.3.dist-info/METADATA,sha256=0VAI5TJKWSFwZriKBYnf5a4MSB5cdOLUh9lV_vYDPJY,3676
-hossam-0.4.3.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-hossam-0.4.3.dist-info/top_level.txt,sha256=_-7bwjhthHplWhywEaHIJX2yL11CQCaLjCNSBlk6wiQ,7
-hossam-0.4.3.dist-info/RECORD,,