statslibx 0.1.6__py3-none-any.whl → 0.1.8__py3-none-any.whl

statslibx/io.py

@@ -0,0 +1,21 @@
+import pandas as pd
+import polars as pl
+from pathlib import Path
+
+
+def load_file(path: str):
+    path = Path(path)
+
+    if not path.exists():
+        raise FileNotFoundError(f"{path} not found")
+
+    if path.suffix == ".csv":
+        return pd.read_csv(path)
+
+    if path.suffix == ".json":
+        return pd.read_json(path)
+
+    if path.suffix in {".txt", ".tsv"}:
+        return pd.read_csv(path, sep="\t")
+
+    raise ValueError(f"Unsupported file type: {path.suffix}")

statslibx/preprocessing/__init__.py

@@ -0,0 +1,228 @@
+from typing import Optional, Union, List, Dict, Any
+import pandas as pd
+import polars as pl
+import numpy as np
+
+
+class Preprocessing:
+
+    def __init__(self, data: Union[pd.DataFrame, pl.DataFrame]):
+        if not isinstance(data, (pd.DataFrame, pl.DataFrame)):
+            raise TypeError("data must be a pandas or polars DataFrame")
+        self.data = data
+
+    # ------------------------------------------------------------------
+    # Internal helpers
+    # ------------------------------------------------------------------
+
+    def _is_pandas(self) -> bool:
+        return isinstance(self.data, pd.DataFrame)
+
+    def _is_polars(self) -> bool:
+        return isinstance(self.data, pl.DataFrame)
+
+    def _count_nulls(self, column: str) -> int:
+        if self._is_pandas():
+            return int(self.data[column].isna().sum())
+        return int(self.data[column].null_count())
+
+    def _get_columns(self, columns):
+        if columns is None:
+            return list(self.data.columns)
+        if isinstance(columns, str):
+            return [columns]
+        return columns
+
+    # ------------------------------------------------------------------
+    # Inspection
+    # ------------------------------------------------------------------
+
+    def detect_nulls(
+        self,
+        columns: Optional[Union[str, List[str]]] = None
+    ) -> pd.DataFrame:
+
+        columns = self._get_columns(columns)
+        total = self.data.shape[0]
+
+        rows = []
+        for col in columns:
+            nulls = self._count_nulls(col)
+            rows.append({
+                "column": col,
+                "nulls": nulls,
+                "non_nulls": total - nulls,
+                "null_pct": nulls / total
+            })
+
+        return pd.DataFrame(rows)
+
+    def check_uniqueness(self) -> pd.DataFrame:
+        if self._is_pandas():
+            unique = self.data.nunique()
+            return pd.DataFrame({
+                "column": unique.index,
+                "unique_values": unique.values
+            })
+
+        unique = self.data.select(pl.all().n_unique())
+        return unique.to_pandas().melt(
+            var_name="column",
+            value_name="unique_values"
+        )
+
+    def preview_data(self, n: int = 5):
+        return self.data.head(n)
+
+    # ------------------------------------------------------------------
+    # Description
+    # ------------------------------------------------------------------
+
+    def describe_numeric(self):
+        if self._is_pandas():
+            return self.data.select_dtypes(include=np.number).describe()
+
+        return self.data.select(pl.all().filter(pl.col(pl.NUMERIC))).describe()
+
+    def describe_categorical(self):
+        if self._is_pandas():
+            return self.data.select_dtypes(include="object").describe()
+
+        return self.data.select(pl.all().filter(pl.col(pl.Utf8))).describe()
+
+    # ------------------------------------------------------------------
+    # Transformations
+    # ------------------------------------------------------------------
+
+    def fill_nulls(
+        self,
+        fill_with: Any,
+        columns: Optional[Union[str, List[str]]] = None
+    ):
+        columns = self._get_columns(columns)
+
+        if self._is_pandas():
+            self.data[columns] = self.data[columns].fillna(fill_with)
+
+        else:
+            self.data = self.data.with_columns([
+                pl.col(col).fill_null(fill_with) for col in columns
+            ])
+
+        return self
+
+    def normalize(self, column: str):
+        if self._is_pandas():
+            col = self.data[column]
+            self.data[column] = (col - col.min()) / (col.max() - col.min())
+        else:
+            self.data = self.data.with_columns(
+                ((pl.col(column) - pl.col(column).min()) /
+                 (pl.col(column).max() - pl.col(column).min()))
+                .alias(column)
+            )
+        return self
+
+    def standardize(self, column: str):
+        if self._is_pandas():
+            col = self.data[column]
+            self.data[column] = (col - col.mean()) / col.std()
+        else:
+            self.data = self.data.with_columns(
+                ((pl.col(column) - pl.col(column).mean()) /
+                 pl.col(column).std())
+                .alias(column)
+            )
+        return self
+
+    # ------------------------------------------------------------------
+    # Filtering
+    # ------------------------------------------------------------------
+
+    def filter_rows(self, condition):
+        if self._is_pandas():
+            self.data = self.data.loc[condition]
+        else:
+            self.data = self.data.filter(condition)
+        return self
+
+    def filter_columns(self, columns: List[str]):
+        if self._is_pandas():
+            self.data = self.data[columns]
+        else:
+            self.data = self.data.select(columns)
+        return self
+
+    def rename_columns(self, mapping: Dict[str, str]):
+        if self._is_pandas():
+            self.data = self.data.rename(columns=mapping)
+        else:
+            self.data = self.data.rename(mapping)
+        return self
+
+    # ------------------------------------------------------------------
+    # Outliers
+    # ------------------------------------------------------------------
+
+    def detect_outliers(
+        self,
+        column: str,
+        method: str = "iqr"
+    ) -> pd.DataFrame:
+        if self._is_pandas():
+            series = self.data[column]
+        else:
+            series = self.data[column].to_pandas()
+
+        # 2. Calcular la máscara según el método
+        if method == "iqr":
+            q1 = series.quantile(0.25)
+            q3 = series.quantile(0.75)
+            iqr = q3 - q1
+            mask_values = (series < q1 - 1.5 * iqr) | (series > q3 + 1.5 * iqr)
+
+        elif method == "zscore":
+            z = (series - series.mean()) / series.std()
+            mask_values = z.abs() > 3
+        else:
+            raise ValueError("method must be 'iqr' or 'zscore'")
+
+        outliers = self.data[mask_values.values]
+
+        # 4. Manejo de retorno profesional
+        if len(outliers) == 0:
+            print(f"No outliers found in column '{column}'")
+            return outliers 
+        
+        return outliers
+
+
+    # ------------------------------------------------------------------
+    # Data Quality Report
+    # ------------------------------------------------------------------
+
+    def data_quality(self) -> pd.DataFrame:
+        total_rows = self.data.shape[0]
+        rows = []
+
+        for col in self.data.columns:
+            nulls = self._count_nulls(col)
+
+            if self._is_pandas():
+                dtype = str(self.data[col].dtype)
+                unique = self.data[col].nunique()
+            else:
+                dtype = str(self.data.schema[col])
+                unique = self.data[col].n_unique()
+
+            rows.append({
+                "column": col,
+                "dtype": dtype,
+                "nulls": nulls,
+                "null_pct": nulls / total_rows,
+                "unique_values": unique,
+                "completeness_pct": 1 - (nulls / total_rows)
+            })
+
+        return pd.DataFrame(rows)
+

statslibx/probability.py

@@ -0,0 +1,2 @@
+class ProbabilityStats:
+    pass

statslibx/utils.py

@@ -398,11 +398,35 @@ class UtilsStats:
         
         return outliers
 
-    def calculate_effect_size(self, group1: np.ndarray, group2: np.ndarray, 
-                                method: Literal['cohen', 'hedges'] = 'cohen') -> dict:
+    def calculate_effect_size(self, 
+                            data: Union[pd.Series, np.ndarray, pd.DataFrame, str, Path] = None,                             
+                            group1: Union[str, pd.Series, np.ndarray] = None, 
+                            group2: Union[str, pd.Series, np.ndarray] = None,
+                            method: Literal['cohen', 'hedges'] = 'cohen') -> dict:
         """
         Calcula el tamaño del efecto entre dos grupos
         """
+
+        # --- Preparar arrays ---
+        # Caso 1: data es DataFrame y group1/group2 son nombres de columna
+        if isinstance(data, pd.DataFrame):
+            group1 = np.array(data[group1])
+            group2 = np.array(data[group2])
+        # Caso 2: data no es None, y es una serie o array, usarlo como group1
+        elif isinstance(data, (pd.Series, np.ndarray)) and group2 is not None:
+            group1 = np.array(data)
+            group2 = np.array(group2)
+        # Caso 3: group1 y group2 ya son arrays o Series
+        else:
+            group1 = np.array(group1)
+            group2 = np.array(group2)
+            
+        # Eliminar nan automáticamente
+        group1 = group1[~np.isnan(group1)]
+        group2 = group2[~np.isnan(group2)]
+
+
+        # --- Calcular estadísticas ---
         mean1, mean2 = np.mean(group1), np.mean(group2)
         std1, std2 = np.std(group1, ddof=1), np.std(group2, ddof=1)
         n1, n2 = len(group1), len(group2)
@@ -434,6 +458,7 @@ class UtilsStats:
             'pooled_std': pooled_std
         }
 
+
     # ============= MÉTODOS DE VISUALIZACIÓN COMPLETOS =============
 
     def _plot_distribution_seaborn(self, data, plot_type, bins, figsize, title, **kwargs):
@@ -528,6 +553,47 @@ class UtilsStats:
             plt.tight_layout()
         
         return fig
+    
+    def _plot_distribution_plotly(self, data, plot_type, bins, title, **kwargs):
+        """Implementación con plotly"""
+        try:
+            import plotly.graph_objects as go
+            import plotly.express as px
+            from plotly.subplots import make_subplots
+        except ImportError:
+            raise ImportError("Plotly no está instalado. Instale con: pip install plotly")
+        
+        if plot_type == 'all':
+            fig = make_subplots(
+                rows=2, cols=2,
+                subplot_titles=('Histograma', 'Box Plot', 'Violin Plot', 'Distribución Acumulada')
+            )
+            
+            # Histograma
+            fig.add_trace(go.Histogram(x=data, nbinsx=bins, name='Histograma'), row=1, col=1)
+            
+            # Box plot
+            fig.add_trace(go.Box(y=data, name='Box Plot'), row=1, col=2)
+            
+            # Violin plot
+            fig.add_trace(go.Violin(y=data, name='Violin Plot'), row=2, col=1)
+            
+            # Distribución acumulada
+            hist, bin_edges = np.histogram(data, bins=bins, density=True)
+            cdf = np.cumsum(hist * np.diff(bin_edges))
+            fig.add_trace(go.Scatter(x=bin_edges[1:], y=cdf, name='CDF'), row=2, col=2)
+            
+        else:
+            if plot_type == 'hist':
+                fig = px.histogram(data, nbins=bins, title=title)
+            elif plot_type == 'box':
+                fig = px.box(y=data, title=title)
+            elif plot_type == 'violin':
+                fig = px.violin(y=data, title=title, box=True)
+            else:
+                fig = px.histogram(data, nbins=bins, title=title)
+        
+        return fig
 
     def plot_distribution(self, 
                             data: Union[pd.DataFrame, pd.Series, np.ndarray, str, Path],
@@ -536,7 +602,7 @@ class UtilsStats:
                             backend: Optional[Literal['matplotlib', 'seaborn', 'plotly']] = "seaborn",
                             bins: int = 30,
                             figsize: Optional[Tuple[int, int]] = None,
-                            save_fig: Optional[bool] = None,
+                            save_fig: Optional[bool] = False,
                             filename: Optional[str] = None,
                             **kwargs):
         """
@@ -568,7 +634,7 @@ class UtilsStats:
         """
         backend = backend or self._plot_backend
         figsize = figsize or self._default_figsize
-        save_fig = save_fig if save_fig is not None else self._save_fig
+        self._save_fig = save_fig
         
         # Resolver datos
         data, source = self._resolve_data(data, column)
@@ -605,59 +671,20 @@ class UtilsStats:
             if save_fig and backend != 'plotly':
                 self._save_figure(fig, filename)
             
-            return fig
+            if backend == 'plotly':
+                return fig
             
         except Exception as e:
             print(f"Error en plot_distribution: {e}")
             raise
 
-    def _plot_distribution_plotly(self, data, plot_type, bins, title, **kwargs):
-        """Implementación con plotly"""
-        try:
-            import plotly.graph_objects as go
-            import plotly.express as px
-            from plotly.subplots import make_subplots
-        except ImportError:
-            raise ImportError("Plotly no está instalado. Instale con: pip install plotly")
-        
-        if plot_type == 'all':
-            fig = make_subplots(
-                rows=2, cols=2,
-                subplot_titles=('Histograma', 'Box Plot', 'Violin Plot', 'Distribución Acumulada')
-            )
-            
-            # Histograma
-            fig.add_trace(go.Histogram(x=data, nbinsx=bins, name='Histograma'), row=1, col=1)
-            
-            # Box plot
-            fig.add_trace(go.Box(y=data, name='Box Plot'), row=1, col=2)
-            
-            # Violin plot
-            fig.add_trace(go.Violin(y=data, name='Violin Plot'), row=2, col=1)
-            
-            # Distribución acumulada
-            hist, bin_edges = np.histogram(data, bins=bins, density=True)
-            cdf = np.cumsum(hist * np.diff(bin_edges))
-            fig.add_trace(go.Scatter(x=bin_edges[1:], y=cdf, name='CDF'), row=2, col=2)
-            
-        else:
-            if plot_type == 'hist':
-                fig = px.histogram(data, nbins=bins, title=title)
-            elif plot_type == 'box':
-                fig = px.box(y=data, title=title)
-            elif plot_type == 'violin':
-                fig = px.violin(y=data, title=title, box=True)
-            else:
-                fig = px.histogram(data, nbins=bins, title=title)
-        
-        return fig
-
     def plot_correlation_matrix(self, 
                                 data: Union[pd.DataFrame, str, Path],
-                                method: str = 'pearson',
-                                backend: Optional[Literal['seaborn', 'plotly']] = None,
+                                method: Literal['pearson', 'kendall', 'spearman'] = 'pearson',
+                                backend: Optional[Literal['seaborn', 'plotly']] = "seaborn",
+                                triangular: Optional[bool] = False,
                                 figsize: Optional[Tuple[int, int]] = None,
-                                save_fig: Optional[bool] = None,
+                                save_fig: Optional[bool] = False,
                                 filename: Optional[str] = None,
                                 **kwargs):
         """
@@ -674,25 +701,32 @@ class UtilsStats:
         """
         backend = backend or self._plot_backend
         figsize = figsize or self._default_figsize
-        save_fig = save_fig if save_fig is not None else self._save_fig
+        self.save_fig = save_fig 
         filename = filename or "matriz_correlacion"
-        
+
         # Resolver datos
         data, source = self._resolve_data(data)
         
         if not isinstance(data, pd.DataFrame):
             raise ValueError("Se requiere un DataFrame para calcular matriz de correlación")
+        else:
+            data = data.select_dtypes(include=['float64', 'int64'])
         
         # Calcular matriz de correlación
         corr_matrix = data.corr(method=method)
         
         if backend == 'seaborn':
             fig, ax = plt.subplots(figsize=figsize)
-            mask = np.triu(np.ones_like(corr_matrix, dtype=bool))
+            if triangular:
+                mask = np.triu(np.ones_like(corr_matrix, dtype=bool))
             
-            sns.heatmap(corr_matrix, mask=mask, annot=True, fmt='.2f', 
-                        cmap='coolwarm', center=0, ax=ax,
-                        square=True, linewidths=0.5, **kwargs)
+                sns.heatmap(corr_matrix, mask=mask, annot=True, fmt='.2f', 
+                            cmap='coolwarm', center=0, ax=ax,
+                            square=True, linewidths=0.5, **kwargs)
+            else:
+                sns.heatmap(corr_matrix, annot=True, fmt='.2f', 
+                            cmap='coolwarm', center=0, ax=ax,
+                            square=True, linewidths=0.5, **kwargs)
             ax.set_title(f'Matriz de Correlación ({method})', fontsize=14, pad=20)
             plt.tight_layout()
             
@@ -731,15 +765,15 @@ class UtilsStats:
                     print(f"✓ Figura Plotly guardada: {filepath}")
                 except Exception as e:
                     print(f"✗ Error guardando figura Plotly: {e}")
-        
-        return fig
+        if backend == 'plotly':
+            return fig
 
     def plot_scatter_matrix(self, 
                             data: Union[pd.DataFrame, str, Path],
                             columns: Optional[List[str]] = None,
                             backend: Optional[Literal['seaborn', 'plotly', 'pandas']] = None,
                             figsize: Optional[Tuple[int, int]] = None,
-                            save_fig: Optional[bool] = None,
+                            save_fig: Optional[bool] = False,
                             filename: Optional[str] = None,
                             **kwargs):
         """
@@ -752,7 +786,7 @@ class UtilsStats:
         """
         backend = backend or self._plot_backend
         figsize = figsize or self._default_figsize
-        save_fig = save_fig if save_fig is not None else self._save_fig
+        self.save_fig = save_fig 
         filename = filename or "scatter_matrix"
         
         # Resolver datos
@@ -791,7 +825,8 @@ class UtilsStats:
                 except Exception as e:
                     print(f"✗ Error guardando figura Plotly: {e}")
         
-        return fig
+        if backend == 'plotly':
+            return fig
 
     # ============= GRÁFICOS CON INTERVALOS DE CONFIANZA =============
 
@@ -802,7 +837,7 @@ class UtilsStats:
                                 ci_method: str = 'parametric',
                                 bins: int = 30,
                                 figsize: Optional[Tuple[int, int]] = None,
-                                save_fig: Optional[bool] = None,
+                                save_fig: Optional[bool] = False,
                                 filename: Optional[str] = None,
                                 **kwargs) -> plt.Figure:
         """
@@ -838,7 +873,7 @@ class UtilsStats:
         x_range = np.linspace(data_array.min(), data_array.max(), 300)
 
         # ======= FIGURA =======
-        fig, (ax1, ax2) = plt.subplots(1, 2, figsize=figsize or (14, 6))
+        fig, (ax1, ax2) = plt.subplots(2, 1, figsize=figsize or (14, 6))
 
         # ============================================================
         # PANEL 1: HISTOGRAMA + KDE
@@ -903,109 +938,37 @@ class UtilsStats:
         plt.tight_layout()
 
         # Guardado opcional
-        save_fig = save_fig if save_fig is not None else self._save_fig
+        self.save_fig = save_fig
         if save_fig:
             self._save_figure(fig, filename)
 
-        return fig
-
-
-    def plot_multiple_distributions_with_ci(self, 
-                                            data_dict: dict,
-                                            confidence_level: float = 0.95,
-                                            figsize: Optional[Tuple[int, int]] = None,
-                                            save_fig: Optional[bool] = None,
-                                            filename: Optional[str] = None,
-                                            **kwargs) -> plt.Figure:
-        """
-        Grafica múltiples distribuciones con sus intervalos de confianza
-        """
-        n_distributions = len(data_dict)
-        fig, axes = plt.subplots(n_distributions, 2, 
-                               figsize=figsize or (14, 5 * n_distributions))
-        
-        if n_distributions == 1:
-            axes = axes.reshape(1, -1)
-        
-        colors = plt.cm.Set3(np.linspace(0, 1, n_distributions))
-        
-        for idx, (name, data) in enumerate(data_dict.items()):
-            ax1, ax2 = axes[idx]
-            
-            if isinstance(data, pd.Series):
-                data_array = data.dropna().values
-            else:
-                data_array = np.array(data)
-                data_array = data_array[~np.isnan(data_array)]
-            
-            # Calcular estadísticas
-            ci_result = self.calculate_confidence_intervals(data_array, confidence_level=confidence_level)
-            
-            # Gráfica izquierda: Distribución básica
-            ax1.hist(data_array, bins=30, alpha=0.7, color=colors[idx], 
-                    edgecolor='black', density=True)
-            
-            kde = stats.gaussian_kde(data_array)
-            x_range = np.linspace(data_array.min(), data_array.max(), 200)
-            ax1.plot(x_range, kde(x_range), 'k-', linewidth=2)
-            ax1.axvline(ci_result['mean'], color='red', linestyle='--', linewidth=2)
-            
-            ax1.set_title(f'{name}\nMedia: {ci_result["mean"]:.2f}')
-            ax1.grid(True, alpha=0.3)
-            
-            # Gráfica derecha: Con intervalos de confianza
-            ax2.hist(data_array, bins=30, alpha=0.7, color=colors[idx], 
-                    edgecolor='black', density=True)
-            ax2.plot(x_range, kde(x_range), 'k-', linewidth=2)
-            
-            ax2.axvline(ci_result['mean'], color='red', linestyle='-', linewidth=3)
-            ax2.axvspan(ci_result['ci_lower'], ci_result['ci_upper'], 
-                        alpha=0.3, color='orange')
-            ax2.axvline(ci_result['ci_lower'], color='orange', linestyle='--', linewidth=2)
-            ax2.axvline(ci_result['ci_upper'], color='orange', linestyle='--', linewidth=2)
-            
-            ax2.set_title(f'{name} con IC {confidence_level*100}%')
-            ax2.grid(True, alpha=0.3)
-        
-        plt.tight_layout()
-        
-        # Guardar figura si está activado
-        save_fig = save_fig if save_fig is not None else self._save_fig
-        if save_fig:
-            filename = filename or "multiples_distribuciones_ci"
-            self._save_figure(fig, filename)
-        
-        return fig
 
     # ============= MÉTODOS UTILITARIOS ADICIONALES =============
 
-    def get_descriptive_stats(self, 
-                                data: Union[pd.DataFrame, pd.Series, np.ndarray, str, Path],
-                                column: Optional[str] = None) -> dict:
-        """
-        Obtiene estadísticas descriptivas completas
-        
-        Ahora acepta rutas de archivos
-        """
-        # Resolver datos
-        data, source = self._resolve_data(data, column)
-        
+    def get_descriptive_stats(self, data, column=None):
+
         if isinstance(data, pd.DataFrame):
             if column is None:
-                raise ValueError("Debe especificar 'column' cuando data es DataFrame")
+                raise ValueError("Debe especificarse una columna")
             data_series = data[column]
-        elif isinstance(data, pd.Series):
-            data_series = data
         else:
             data_series = pd.Series(data)
-        
+
         data_clean = data_series.dropna()
-        
+
+        if len(data_clean) == 0:
+            return {k: np.nan for k in [
+                'count','mean','median','mode','std','variance',
+                'min','max','q1','q3','iqr','skewness','kurtosis','range'
+            ]}
+
+        mode_result = stats.mode(data_clean, keepdims=False)
+
         return {
             'count': len(data_clean),
             'mean': np.mean(data_clean),
             'median': np.median(data_clean),
-            'mode': stats.mode(data_clean)[0][0] if len(data_clean) > 0 else np.nan,
+            'mode': mode_result.mode,
             'std': np.std(data_clean, ddof=1),
             'variance': np.var(data_clean, ddof=1),
             'min': np.min(data_clean),
@@ -1017,7 +980,6 @@ class UtilsStats:
             'kurtosis': stats.kurtosis(data_clean),
             'range': np.max(data_clean) - np.min(data_clean)
         }
-
     def help(self):
         """
         Muestra ayuda completa de la clase DescriptiveStats