davis-stats 1.0__tar.gz

davis_stats-1.0/PKG-INFO

@@ -0,0 +1,21 @@
+Metadata-Version: 2.4
+Name: davis_stats
+Version: 1.0
+Summary: davis_stats is a teaching-focused repository 
+Author: Justin G. Davis
+Author-email: 
+Keywords: statistics,data science,business analytics
+Requires-Python: >=3.7
+Requires-Dist: pandas>=1.0.0
+Requires-Dist: openpyxl>=3.0.0
+Requires-Dist: numpy>=1.20.0
+Requires-Dist: matplotlib>=3.0.0
+Requires-Dist: scipy>=1.6.0
+Requires-Dist: seaborn>=0.12.0
+Requires-Dist: statsmodels>=0.14.0
+Requires-Dist: ipympl>=0.8.0
+Dynamic: author
+Dynamic: keywords
+Dynamic: requires-dist
+Dynamic: requires-python
+Dynamic: summary

davis_stats-1.0/davis_stats/__init__.py

@@ -0,0 +1,6 @@
+from .visualization.trim import trim
+from .visualization.boxplot import boxplot
+from .visualization.histogram import histogram
+from .visualization.scatter import scatter
+from .data import ceo_comp
+from .stats.reg import reg

davis_stats-1.0/davis_stats/data/__init__.py

@@ -0,0 +1,18 @@
+import pandas as pd
+from pathlib import Path
+
+DATA_DIR = Path(__file__).parent
+
+class DataFrames:
+    _ceo_comp = None
+    
+    @property
+    def ceo_comp(self):
+        if self._ceo_comp is None:
+            self._ceo_comp = pd.read_excel(DATA_DIR / 'ceo_comp.xlsx')
+        return self._ceo_comp
+
+_data = DataFrames()
+
+def ceo_comp():
+    return _data.ceo_comp

davis_stats-1.0/davis_stats/stats/__init__.py

@@ -0,0 +1 @@
+from .reg import reg

davis_stats-1.0/davis_stats/stats/reg.py

@@ -0,0 +1,110 @@
+def reg(df, y, x, dummies=None, logistic=False):
+    """
+    Run linear (OLS) or logistic regression.
+    
+    Parameters
+    ----------
+    df : pd.DataFrame
+        Input dataframe containing all variables
+    y : str
+        Name of the dependent variable column
+    x : str or list of str
+        Name(s) of independent variable column(s)
+    dummies : str or list of str, optional
+        Categorical variable(s) to convert to dummy variables
+    logistic : bool, default False
+        If True, run logistic regression; otherwise run OLS
+    
+    Returns
+    -------
+    statsmodels results object or None if fitting fails
+    """
+    import statsmodels.api as sm
+    import pandas as pd
+    
+    # Convert x to list if string, make copy to avoid modifying original
+    if isinstance(x, str):
+        x = [x]
+    else:
+        x = list(x)
+    
+    df_reg = df.copy()
+    
+    # Convert main variables to numeric
+    for col in [y] + x:
+        df_reg[col] = pd.to_numeric(df_reg[col], errors='coerce')
+    
+    # Handle dummy variables
+    if dummies:
+        if isinstance(dummies, str):
+            dummies = [dummies]
+        
+        for dummy_var in dummies:
+            # For logistic: filter to categories with variation in y
+            if logistic:
+                ct = pd.crosstab(df_reg[dummy_var], df_reg[y])
+                valid_categories = ct[(ct > 0).all(axis=1)].index
+                df_reg = df_reg[df_reg[dummy_var].isin(valid_categories)]
+            
+            dummy_cols = pd.get_dummies(
+                df_reg[dummy_var], 
+                prefix=dummy_var, 
+                drop_first=True, 
+                dtype=float
+            )
+            df_reg = pd.concat([df_reg, dummy_cols], axis=1)
+            x.extend(dummy_cols.columns.tolist())
+    
+    # Drop rows with missing values
+    df_reg = df_reg.dropna(subset=[y] + x)
+    
+    if len(df_reg) == 0:
+        print("Error: No observations remaining after dropping missing values")
+        return None
+    
+    # Prepare X and y
+    X = sm.add_constant(df_reg[x].astype(float))
+    y_data = df_reg[y].astype(float)
+    
+    # Fit model
+    try:
+        if logistic:
+            model = sm.Logit(y_data, X)
+            results = model.fit(method='bfgs', maxiter=100, disp=0)
+        else:
+            model = sm.OLS(y_data, X)
+            results = model.fit()
+        
+        print(results.summary())
+        return results
+        
+    except Exception as e:
+        print(f"Error fitting model: {e}")
+        
+        # For logistic with dummies, try again with larger categories only
+        if logistic and dummies:
+            min_obs = 30
+            print(f"\nRetrying with categories having at least {min_obs} observations...")
+            
+            for dummy_var in dummies:
+                value_counts = df_reg[dummy_var].value_counts()
+                valid_categories = value_counts[value_counts >= min_obs].index
+                df_reg = df_reg[df_reg[dummy_var].isin(valid_categories)]
+            
+            if len(df_reg) == 0:
+                print("Error: No observations remaining after filtering")
+                return None
+            
+            try:
+                X = sm.add_constant(df_reg[x].astype(float))
+                y_data = df_reg[y].astype(float)
+                
+                model = sm.Logit(y_data, X)
+                results = model.fit(method='bfgs', maxiter=100, disp=0)
+                print("\nResults with reduced sample:")
+                print(results.summary())
+                return results
+            except Exception as e2:
+                print(f"Error fitting reduced model: {e2}")
+        
+        return None

davis_stats-1.0/davis_stats/visualization/__init__.py

@@ -0,0 +1,4 @@
+from .trim import trim
+from .boxplot import boxplot
+from .histogram import histogram
+from .scatter import scatter

davis_stats-1.0/davis_stats/visualization/boxplot.py

@@ -0,0 +1,34 @@
+import matplotlib.pyplot as plt
+import numpy as np
+from .trim import trim
+
+def boxplot(series, title=None, trim_outliers=100, dpi=150, figsize=(6, 4)):
+    # Create figure and axis
+    fig, ax = plt.subplots(figsize=figsize, dpi=dpi)
+    
+    if not title:
+        title = series.name
+        
+    # Apply trimming if trim < 100
+    if trim_outliers < 100:
+        series = trim(series, trim_outliers)
+        title = f"{title} (outliers removed at {trim_outliers}% level)"
+    
+    # Convert to numpy array and ensure 1D
+    data = np.array(series.dropna()).flatten()
+    
+    # Create boxplot on the axis
+    ax.boxplot(data,
+        patch_artist=True,
+        boxprops=dict(facecolor='skyblue', color='black'),
+        medianprops=dict(color='black'),
+        flierprops=dict(marker='o', markerfacecolor='gray'),
+        whiskerprops=dict(color='black'),
+        capprops=dict(color='black'))
+    
+    # Add labels and styling
+    ax.set_title(title)
+    ax.ticklabel_format(style='plain', axis='y')
+    ax.grid(True, linestyle='--', alpha=0.7)
+    
+    plt.show(block=False)

davis_stats-1.0/davis_stats/visualization/histogram.py

@@ -0,0 +1,71 @@
+import matplotlib.pyplot as plt
+import numpy as np
+import pandas as pd
+from .trim import trim
+
+def histogram(series, title=None, bins=30, trim_outliers=100, details=False, dpi=150, figsize=(6, 4)):
+    # Create figure and axis
+    fig, ax = plt.subplots(figsize=figsize, dpi=dpi)
+    
+    # Get base title
+    if not title:
+        title = series.name
+    
+    # Store original data for statistics BEFORE any trimming
+    original_data = np.array(series.dropna()).flatten()
+    
+    # Apply trimming if trim < 100
+    if trim_outliers < 100:
+        series = trim(series, trim_outliers)
+        title = f"{title} (outliers removed at {trim_outliers}% level)"
+    
+    # Convert to numpy array and handle NaN
+    data = np.array(series.dropna()).flatten()
+    
+    # Create histogram
+    ax.hist(data, 
+            bins=bins,
+            edgecolor='black',
+            color='skyblue',
+            alpha=0.7)
+    
+    # Add labels and styling
+    ax.set_title(title)
+    ax.set_ylabel('Count')
+    ax.ticklabel_format(style='plain', axis='x')
+    ax.grid(True, linestyle='--', alpha=0.7)
+    
+    if details:
+        # Add mean and median lines
+        mean = np.mean(original_data)
+        median = np.median(original_data)
+        ax.axvline(mean, color='red', linestyle='--', label=f'Mean: {mean:.2f}')
+        ax.axvline(median, color='green', linestyle='--', label=f'Median: {median:.2f}')
+        
+        # Calculate standard deviations
+        std = np.std(original_data)
+        
+        # Calculate percentages within SDs
+        within_1sd = np.sum((original_data >= mean - std) & (original_data <= mean + std)) / len(original_data) * 100
+        within_2sd = np.sum((original_data >= mean - 2*std) & (original_data <= mean + 2*std)) / len(original_data) * 100
+        within_3sd = np.sum((original_data >= mean - 3*std) & (original_data <= mean + 3*std)) / len(original_data) * 100
+        
+        skew_val = pd.Series(original_data).skew()
+        
+        # Add text box with statistics
+        stats_text = (
+            f'Skewness: {skew_val:.3f}\n'
+            f'Within 1 SD: {within_1sd:.1f}%\n'
+            f'Within 2 SD: {within_2sd:.1f}%\n'
+            f'Within 3 SD: {within_3sd:.1f}%'
+        )
+        plt.text(0.95, 0.95, stats_text,
+                transform=ax.transAxes,
+                verticalalignment='top',
+                horizontalalignment='right',
+                bbox=dict(boxstyle='round', facecolor='white', alpha=0.8))
+        
+        # Add legend
+        plt.legend()
+    
+    plt.show(block=False)

davis_stats-1.0/davis_stats/visualization/scatter.py

@@ -0,0 +1,132 @@
+def scatter(df, y, x, z=None, fit_line=False, dpi=150, figsize=(6, 4)):
+    """
+    Create a nice scatter plot with optional fit line and correlation coefficient
+    
+    Parameters:
+    df (pandas DataFrame): Input data
+    y (str): Column name for y-axis variable (vertical axis in 3D)
+    x (str): Column name for x-axis variable
+    z (str, optional): Column name for z-axis variable (creates 3D plot)
+    fit_line (bool): If True, adds best fit line (2D) or plane (3D)
+    dpi (int): Plot resolution
+    figsize (tuple): Figure size
+    """
+    import seaborn as sns
+    import matplotlib.pyplot as plt
+    import numpy as np
+    from mpl_toolkits.mplot3d import Axes3D
+    
+    # 2D scatter plot (original functionality)
+    if z is None:
+        # Calculate correlation coefficient
+        corr = df[x].corr(df[y])
+        
+        # Set style
+        sns.set_style("whitegrid")
+        
+        # Create figure
+        fig, ax = plt.subplots(figsize=figsize, dpi=dpi)
+        
+        # Create scatter plot
+        if fit_line:
+            # Use seaborn's regplot for scatter + fit line
+            sns.regplot(data=df, 
+                       x=x, 
+                       y=y,
+                       scatter_kws={'alpha':0.5},
+                       line_kws={'color': 'red'},
+                       ci=None)
+        else:
+            # Use seaborn's scatterplot
+            sns.scatterplot(data=df,
+                           x=x,
+                           y=y,
+                           alpha=0.5)
+        
+        # Customize plot - 2 variable title format
+        plt.title(f'{y} and {x}\nCorrelation: {corr:.3f}', pad=15)
+        plt.xlabel(x)
+        plt.ylabel(y)
+        
+        # Adjust layout
+        plt.tight_layout()
+        
+    # 3D scatter plot
+    else:
+        # Create 3D figure with larger size for better visibility
+        fig = plt.figure(figsize=figsize, dpi=dpi)
+        ax = fig.add_subplot(111, projection='3d')
+        
+        # Remove NaN values for plotting
+        plot_df = df[[x, y, z]].dropna()
+        
+        # Create 3D scatter plot with color spectrum and edge color
+        # In matplotlib 3D: (x-axis, y-axis, z-axis) where z-axis is VERTICAL
+        # So to make y variable vertical, it goes in the 3rd position
+        scatter = ax.scatter(plot_df[x], plot_df[z], plot_df[y], 
+                            c=plot_df[y], cmap='RdYlBu_r', s=50,
+                            edgecolor='black', linewidth=0.5,
+                            alpha=0.8)
+        
+        # Add colorbar with more space
+        cbar = plt.colorbar(scatter, ax=ax, pad=0.15, shrink=0.8)
+        cbar.set_label(y, rotation=270, labelpad=15)
+        
+        # Add best-fit plane if requested
+        if fit_line:
+            # Prepare data for plane fitting: y = a*x + b*z + c
+            X_data = np.column_stack([plot_df[x], plot_df[z], np.ones(len(plot_df))])
+            y_data = plot_df[y].values
+            
+            # Fit plane using least squares: [a, b, c]
+            coeffs, residuals, rank, s = np.linalg.lstsq(X_data, y_data, rcond=None)
+            a, b, c = coeffs
+            
+            # Create mesh grid for the regression plane
+            x_surf = np.linspace(plot_df[x].min(), plot_df[x].max(), 20)
+            z_surf = np.linspace(plot_df[z].min(), plot_df[z].max(), 20)
+            X_mesh, Z_mesh = np.meshgrid(x_surf, z_surf)
+            
+            # Calculate y values for the plane
+            Y_mesh = a * X_mesh + b * Z_mesh + c
+            
+            # Plot the regression plane with grid lines and semi-transparency
+            # Order: (x-axis, y-axis, z-axis) where z-axis is vertical
+            surf = ax.plot_surface(X_mesh, Z_mesh, Y_mesh, 
+                                   alpha=0.4, cmap='coolwarm',
+                                   edgecolor='black', linewidth=0.5,
+                                   rstride=1, cstride=1,
+                                   antialiased=True)
+        
+        # 3 variable title format
+        title = f'{y}, {x}, and {z}'
+        
+        # Set axis labels
+        # In matplotlib 3D: x is horizontal left-right, y is horizontal front-back, z is VERTICAL
+        ax.set_xlabel(x, labelpad=10)      # horizontal axis
+        ax.set_ylabel(z, labelpad=10)      # horizontal axis (front-back)
+        ax.set_zlabel(y, labelpad=10)      # VERTICAL axis (up-down)
+        
+        # Move z-axis (vertical) ticks and label to the left to avoid colorbar overlap
+        ax.zaxis._axinfo['juggled'] = (1, 2, 0)  # Move z-axis to left side
+        
+        # Flip axes to go from least to greatest (normal order)
+        ax.invert_xaxis()
+        ax.invert_yaxis()
+        
+        # Adjust the viewing angle for better perspective
+        ax.view_init(elev=25, azim=135)
+        
+        # Make the grid lines more visible
+        ax.xaxis._axinfo["grid"]['color'] = (0.5, 0.5, 0.5, 0.5)
+        ax.yaxis._axinfo["grid"]['color'] = (0.5, 0.5, 0.5, 0.5)
+        ax.zaxis._axinfo["grid"]['color'] = (0.5, 0.5, 0.5, 0.5)
+        
+        # Set title after everything else
+        ax.set_title(title, pad=15)
+        
+        # Adjust layout
+        plt.tight_layout()
+    
+    # Show plot
+    plt.show()

davis_stats-1.0/davis_stats/visualization/trim.py

@@ -0,0 +1,16 @@
+import numpy as np
+
+def trim(series, percentile_keep = 100):
+    # Calculate how much to remove
+    remove_pct = 100 - percentile_keep
+    
+    # Find the threshold for most extreme values
+    # This gets the absolute distance from median for each point
+    median_val = np.median(series)
+    abs_deviations = np.abs(series - median_val)
+    
+    # Find the threshold - keep values with smaller deviations
+    threshold = np.percentile(abs_deviations, percentile_keep)
+    
+    # Return values within the threshold
+    return series[abs_deviations <= threshold]

davis_stats-1.0/davis_stats.egg-info/PKG-INFO

@@ -0,0 +1,21 @@
+Metadata-Version: 2.4
+Name: davis_stats
+Version: 1.0
+Summary: davis_stats is a teaching-focused repository 
+Author: Justin G. Davis
+Author-email: 
+Keywords: statistics,data science,business analytics
+Requires-Python: >=3.7
+Requires-Dist: pandas>=1.0.0
+Requires-Dist: openpyxl>=3.0.0
+Requires-Dist: numpy>=1.20.0
+Requires-Dist: matplotlib>=3.0.0
+Requires-Dist: scipy>=1.6.0
+Requires-Dist: seaborn>=0.12.0
+Requires-Dist: statsmodels>=0.14.0
+Requires-Dist: ipympl>=0.8.0
+Dynamic: author
+Dynamic: keywords
+Dynamic: requires-dist
+Dynamic: requires-python
+Dynamic: summary

davis_stats-1.0/davis_stats.egg-info/SOURCES.txt

@@ -0,0 +1,16 @@
+setup.py
+davis_stats/__init__.py
+davis_stats.egg-info/PKG-INFO
+davis_stats.egg-info/SOURCES.txt
+davis_stats.egg-info/dependency_links.txt
+davis_stats.egg-info/requires.txt
+davis_stats.egg-info/top_level.txt
+davis_stats/data/__init__.py
+davis_stats/data/ceo_comp.xlsx
+davis_stats/stats/__init__.py
+davis_stats/stats/reg.py
+davis_stats/visualization/__init__.py
+davis_stats/visualization/boxplot.py
+davis_stats/visualization/histogram.py
+davis_stats/visualization/scatter.py
+davis_stats/visualization/trim.py

davis_stats-1.0/davis_stats.egg-info/dependency_links.txt

@@ -0,0 +1 @@
+

davis_stats-1.0/davis_stats.egg-info/requires.txt

@@ -0,0 +1,8 @@
+pandas>=1.0.0
+openpyxl>=3.0.0
+numpy>=1.20.0
+matplotlib>=3.0.0
+scipy>=1.6.0
+seaborn>=0.12.0
+statsmodels>=0.14.0
+ipympl>=0.8.0

davis_stats-1.0/davis_stats.egg-info/top_level.txt

@@ -0,0 +1 @@
+davis_stats

davis_stats-1.0/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

davis_stats-1.0/setup.py

@@ -0,0 +1,28 @@
+from setuptools import setup, find_packages
+
+setup(
+    name='davis_stats',
+    version='1.0',
+    packages=find_packages(),
+    package_data={
+        'davis_stats': ['data/*.xlsx']},
+    install_requires=[
+        'pandas>=1.0.0',
+        'openpyxl>=3.0.0',
+        'numpy>=1.20.0',
+        'matplotlib>=3.0.0',
+        'scipy>=1.6.0',
+        'seaborn>=0.12.0',
+        'statsmodels>=0.14.0',
+        'ipympl>=0.8.0'],
+    python_requires='>=3.7',
+    author='Justin G. Davis',
+    author_email='',
+    description='''
+        davis_stats is a teaching-focused repository 
+        for applied statistics, data science, and 
+        business analytics. It contains functions 
+        and datasets to help W&L students develop 
+        practical skills in data analysis, statistical 
+        modeling, and real-world decision making.''',
+    keywords='statistics, data science, business analytics')