well-log-toolkit 0.1.118__tar.gz → 0.1.119__tar.gz

{well_log_toolkit-0.1.118 → well_log_toolkit-0.1.119}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: well-log-toolkit
-Version: 0.1.118
+Version: 0.1.119
 Summary: Fast LAS file processing with lazy loading and filtering for well log analysis
 Author-email: Kristian dF Kollsgård <kkollsg@gmail.com>
 License: MIT

{well_log_toolkit-0.1.118 → well_log_toolkit-0.1.119}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "well-log-toolkit"
-version = "0.1.118"
+version = "0.1.119"
 description = "Fast LAS file processing with lazy loading and filtering for well log analysis"
 readme = "README.md"
 requires-python = ">=3.9"

{well_log_toolkit-0.1.118 → well_log_toolkit-0.1.119}/well_log_toolkit/regression.py

@@ -70,13 +70,14 @@ class RegressionBase(ABC):
         """
         return self.predict(x)
 
-    def _calculate_metrics(self, x: np.ndarray, y: np.ndarray, y_pred: np.ndarray) -> None:
+    def _calculate_metrics(self, x: np.ndarray, y: np.ndarray, y_pred: np.ndarray, use_log_space: bool = False) -> None:
         """Calculate R² and RMSE metrics.
 
         Args:
             x: Independent variable values
             y: Actual dependent variable values
             y_pred: Predicted dependent variable values
+            use_log_space: If True, calculate R² in log space (useful when y spans orders of magnitude)
         """
         # Store original data
         self.x_data = x
@@ -85,12 +86,21 @@ class RegressionBase(ABC):
         # Store x-axis range
         self.x_range = (float(np.min(x)), float(np.max(x)))
 
-        # R² calculation
-        ss_res = np.sum((y - y_pred) ** 2)
-        ss_tot = np.sum((y - np.mean(y)) ** 2)
-        self.r_squared = 1 - (ss_res / ss_tot) if ss_tot != 0 else 0.0
+        # R² calculation - in log space if requested
+        if use_log_space and np.all(y > 0) and np.all(y_pred > 0):
+            # Calculate R² in log space for data spanning orders of magnitude
+            log_y = np.log10(y)
+            log_y_pred = np.log10(y_pred)
+            ss_res = np.sum((log_y - log_y_pred) ** 2)
+            ss_tot = np.sum((log_y - np.mean(log_y)) ** 2)
+            self.r_squared = 1 - (ss_res / ss_tot) if ss_tot != 0 else 0.0
+        else:
+            # Standard R² in linear space
+            ss_res = np.sum((y - y_pred) ** 2)
+            ss_tot = np.sum((y - np.mean(y)) ** 2)
+            self.r_squared = 1 - (ss_res / ss_tot) if ss_tot != 0 else 0.0
 
-        # RMSE calculation
+        # RMSE calculation (always in linear space)
         self.rmse = np.sqrt(np.mean((y - y_pred) ** 2))
 
     def _prepare_data(self, x: ArrayLike, y: ArrayLike) -> Tuple[np.ndarray, np.ndarray]:

{well_log_toolkit-0.1.118 → well_log_toolkit-0.1.119}/well_log_toolkit/visualization.py

@@ -3139,6 +3139,27 @@ class Crossplot:
             self._regressions[reg_type] = {}
         self._regressions[reg_type][identifier] = regression_obj
 
+    def _get_group_colors(self, data: pd.DataFrame, group_column: str) -> dict:
+        """Get the color assigned to each group in the plot.
+
+        Args:
+            data: DataFrame with plotting data
+            group_column: Column name used for grouping
+
+        Returns:
+            Dictionary mapping group names to their colors
+        """
+        group_colors = {}
+
+        # Get unique groups in the same order as they'll appear in the plot
+        groups = data.groupby(group_column)
+
+        for idx, (group_name, _) in enumerate(groups):
+            # Use the same color assignment logic as _plot_by_groups
+            group_colors[group_name] = DEFAULT_COLORS[idx % len(DEFAULT_COLORS)]
+
+        return group_colors
+
     def _find_best_legend_locations(self, data: pd.DataFrame) -> tuple[str, str]:
         """Find the two best locations for legends based on data density.
 
@@ -3228,7 +3249,12 @@ class Crossplot:
 
         # Add R² on second line if requested (will be styled grey later)
         if include_r2:
-            return f"{first_line}\nR² = {reg.r_squared:.3f}"
+            # Format R² with appropriate note
+            if self.y_log:
+                r2_label = f"R² = {reg.r_squared:.3f} (log)"
+            else:
+                r2_label = f"R² = {reg.r_squared:.3f}"
+            return f"{first_line}\n{r2_label}"
         else:
             return first_line
 
@@ -3371,15 +3397,19 @@ class Crossplot:
                             f"Reduce the number of groups or use a different regression strategy."
                         )
 
+                    # Get the actual colors used for each group in the plot
+                    group_colors_map = self._get_group_colors(data, group_column)
+
                     for idx, (group_name, group_data) in enumerate(color_groups):
                         x_vals = group_data['x'].values
                         y_vals = group_data['y'].values
                         mask = np.isfinite(x_vals) & np.isfinite(y_vals)
                         if np.sum(mask) >= 2:
-                            # Copy config and set default line color if not specified
+                            # Copy config and use the same color as the data group
                             group_config = config.copy()
                             if 'line_color' not in group_config:
-                                group_config['line_color'] = regression_colors[color_idx % len(regression_colors)]
+                                # Use the same color as the data points for this group
+                                group_config['line_color'] = group_colors_map.get(group_name, regression_colors[color_idx % len(regression_colors)])
 
                             # Skip legend update for all but last regression
                             is_last = (idx == n_groups - 1)
@@ -3412,15 +3442,19 @@ class Crossplot:
                         f"Reduce the number of groups or use a different regression strategy."
                     )
 
+                # Get the actual colors used for each group in the plot
+                group_colors_map = self._get_group_colors(data, group_col)
+
                 for idx, (group_name, group_data) in enumerate(groups):
                     x_vals = group_data['x'].values
                     y_vals = group_data['y'].values
                     mask = np.isfinite(x_vals) & np.isfinite(y_vals)
                     if np.sum(mask) >= 2:
-                        # Copy config and set default line color if not specified
+                        # Copy config and use the same color as the data group
                         group_config = config.copy()
                         if 'line_color' not in group_config:
-                            group_config['line_color'] = regression_colors[color_idx % len(regression_colors)]
+                            # Use the same color as the data points for this group
+                            group_config['line_color'] = group_colors_map.get(group_name, regression_colors[color_idx % len(regression_colors)])
 
                         # Skip legend update for all but last regression
                         is_last = (idx == n_groups - 1)
@@ -3467,6 +3501,11 @@ class Crossplot:
             warnings.warn(f"Failed to fit {regression_type} regression for {name}: {e}")
             return
 
+        # Recalculate R² in log space if y-axis is log scale
+        if self.y_log:
+            y_pred = reg.predict(x_vals)
+            reg._calculate_metrics(x_vals, y_vals, y_pred, use_log_space=True)
+
         # Store regression in nested structure
         self._store_regression(regression_type, name, reg)
 
@@ -3810,6 +3849,11 @@ class Crossplot:
         except ValueError as e:
             raise ValueError(f"Failed to fit {regression_type} regression: {e}")
 
+        # Recalculate R² in log space if y-axis is log scale
+        if self.y_log:
+            y_pred = reg.predict(x_clean)
+            reg._calculate_metrics(x_clean, y_clean, y_pred, use_log_space=True)
+
         # Store regression in nested structure
         reg_name = name if name else regression_type
         self._store_regression(regression_type, reg_name, reg)

{well_log_toolkit-0.1.118 → well_log_toolkit-0.1.119}/well_log_toolkit.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: well-log-toolkit
-Version: 0.1.118
+Version: 0.1.119
 Summary: Fast LAS file processing with lazy loading and filtering for well log analysis
 Author-email: Kristian dF Kollsgård <kkollsg@gmail.com>
 License: MIT