Delete Utils

utils/helpers.py

@@ -1,28 +0,0 @@
-# causalscience/utils/helpers.py
-
-import pandas as pd
-
-
-def detect_column_type(df: pd.DataFrame, column: str) -> str:
-    """
-    Determine if a column is boolean (binary) or timeseries.
-
-    Args:
-        df (pd.DataFrame): Input DataFrame.
-        column (str): Column name to inspect.
-
-    Returns:
-        'boolean' if the column has exactly two unique values,
-        'timeseries' if all values can be parsed as dates,
-        None otherwise.
-    """
-    unique_vals = df[column].dropna().unique()
-    if len(unique_vals) == 2:
-        return 'boolean'
-    try:
-        dt = pd.to_datetime(df[column], errors='coerce')
-        if dt.notna().all():
-            return 'timeseries'
-    except Exception:
-        pass
-    return None

utils/plotting.py

@@ -1,77 +0,0 @@
-# causalscience/utils/plotting.py
-
-import numpy as np
-import matplotlib.pyplot as plt
-from io import BytesIO
-from PIL import Image
-
-
-def calculate_standardized_differences(data, vars_list, treatment, weights=None):
-    """
-    Calculate standardized mean differences for covariate balance.
-
-    Args:
-        data (pd.DataFrame): Dataset including treatment indicator.
-        vars_list (list[str]): Numeric covariate column names.
-        treatment (str): Treatment indicator column (0/1).
-        weights (list or np.array, optional): Weights for each observation.
-
-    Returns:
-        pd.DataFrame: Columns ['variable', 'std_diff'].
-    """
-    results = []
-    if weights is None:
-        weights = np.ones(len(data))
-    else:
-        weights = np.array(weights)
-
-    treated = data[treatment] == 1
-    control = data[treatment] == 0
-
-    for var in vars_list:
-        mean_t = np.average(data.loc[treated, var], weights=weights[treated])
-        mean_c = np.average(data.loc[control, var], weights=weights[control])
-        var_t = np.average((data.loc[treated, var] - mean_t)**2, weights=weights[treated])
-        var_c = np.average((data.loc[control, var] - mean_c)**2, weights=weights[control])
-        pooled_sd = np.sqrt((var_t + var_c) / 2)
-        std_diff = (mean_t - mean_c) / pooled_sd if pooled_sd != 0 else np.nan
-        results.append({'variable': var, 'std_diff': std_diff})
-    import pandas as pd
-    return pd.DataFrame(results)
-
-
-def love_plot(std_diffs_list, labels, threshold=0.1, abs_val=False):
-    """
-    Generate a Love plot for covariate balance.
-
-    Args:
-        std_diffs_list (list[pd.DataFrame]): List of std diff DataFrames.
-        labels (list[str]): Labels for each dataset.
-        threshold (float, optional): Threshold lines for balance.
-        abs_val (bool, optional): Plot absolute std diffs.
-
-    Returns:
-        PIL.Image.Image: Love plot saved to image buffer.
-    """
-    fig, ax = plt.subplots(figsize=(10, 6))
-    markers = ['o', 's', '^', 'd']
-    colors = ['red', 'blue', 'green', 'purple']
-
-    for i, (std_df, label) in enumerate(zip(std_diffs_list, labels)):
-        values = std_df['std_diff'].abs() if abs_val else std_df['std_diff']
-        ax.scatter(values, std_df['variable'], label=label,
-                   marker=markers[i % len(markers)], s=100)
-
-    ax.axvline(x=threshold, color='gray', linestyle='--', alpha=0.5)
-    ax.axvline(x=-threshold, color='gray', linestyle='--', alpha=0.5)
-
-    ax.set_xlabel('Standardized Mean Difference')
-    ax.set_title('Love Plot: Covariate Balance')
-    ax.legend()
-
-    buf = BytesIO()
-    fig.tight_layout()
-    fig.savefig(buf, format='png', bbox_inches='tight')
-    plt.close(fig)
-    buf.seek(0)
-    return Image.open(buf)