Merge pull request #281 from uriahf/273-small-changes-for-better-calibration-plots

273 small changes for better calibration plots

Author: uriahf

pyproject.toml

@@ -15,7 +15,7 @@ dependencies = [
     "polars>=1.31.0",
 ]
 name = "rtichoke"
-version = "0.1.27"
+version = "0.1.28"
 description = "interactive visualizations for performance of predictive models"
 readme = "README.md"
 

src/rtichoke/calibration/calibration.py

@@ -2,7 +2,7 @@
 A module for Calibration Curves
 """
 
-from typing import Any, Dict, List, Union
+from typing import Any, Dict, List, Union, cast
 
 # import pandas as pd
 import plotly.graph_objects as go
@@ -247,6 +247,7 @@ def _create_plotly_curve_from_calibration_curve_list_times(
         },
         barmode="overlay",
         plot_bgcolor="rgba(0, 0, 0, 0)",
+        paper_bgcolor="rgba(0, 0, 0, 0)",
         legend={
             "orientation": "h",
             "xanchor": "center",
@@ -285,6 +286,7 @@ def _create_plotly_curve_from_calibration_curve_list(
             "yaxis": {"showgrid": False},
             "barmode": "overlay",
             "plot_bgcolor": "rgba(0, 0, 0, 0)",
+            "paper_bgcolor": "rgba(0, 0, 0, 0)",
             "legend": {
                 "orientation": "h",
                 "xanchor": "center",
@@ -470,7 +472,7 @@ def _make_deciles_dat_binary(
     if isinstance(reals, dict):
         reference_groups_keys = list(reals.keys())
         y_list = [
-            np.asarray(reals[reference_group]).ravel()
+            np.asarray(reals[str(reference_group)]).ravel()
             for reference_group in reference_groups_keys
         ]
         lengths = np.array([len(y) for y in y_list], dtype=np.int64)
@@ -533,7 +535,7 @@ def _make_deciles_dat_binary(
             (
                 (pl.col("prob").rank("ordinal").over(["reference_group", "model"]) - 1)
                 * n_bins
-                // pl.count().over(["reference_group", "model"])
+                // pl.len().over(["reference_group", "model"])
                 + 1
             ).alias("decile"),
         ]
@@ -602,7 +604,7 @@ def _create_calibration_curve_list(
 
     reference_data = _create_reference_data_for_calibration_curve()
 
-    reference_groups = deciles_data["reference_group"].unique().to_list()
+    reference_groups = list(probs.keys())
 
     colors_dictionary = _create_colors_dictionary_for_calibration(
         reference_groups, color_values, performance_type
@@ -689,7 +691,9 @@ def process_single_array(p, r, group_name):
                 for group_name in reals.keys():
                     if group_name in probs:
                         frame = process_single_array(
-                            probs[group_name], reals[group_name], group_name
+                            probs[str(group_name)],
+                            reals[str(group_name)],
+                            str(group_name),
                         )
                         smooth_frames.append(frame)
 
@@ -856,8 +860,21 @@ def _define_limits_for_calibration_plot(deciles_dat: pl.DataFrame) -> List[float
     if deciles_dat.height == 1:
         lower_bound, upper_bound = 0.0, 1.0
     else:
-        lower_bound = float(max(0, min(deciles_dat["x"].min(), deciles_dat["y"].min())))
-        upper_bound = float(max(deciles_dat["x"].max(), deciles_dat["y"].max()))
+        lower_bound = float(
+            max(
+                0,
+                min(
+                    cast(float, deciles_dat["x"].min()),
+                    cast(float, deciles_dat["y"].min()),
+                ),
+            )
+        )
+        upper_bound = float(
+            max(
+                cast(float, deciles_dat["x"].max()),
+                cast(float, deciles_dat["y"].max()),
+            )
+        )
 
     return [
         lower_bound - (upper_bound - lower_bound) * 0.05,
@@ -1101,7 +1118,7 @@ def _create_calibration_curve_list_times(
     )
 
     reference_data = _create_reference_data_for_calibration_curve()
-    reference_groups = deciles_dat_final["reference_group"].unique().to_list()
+    reference_groups = list(probs.keys())
     colors_dictionary = _create_colors_dictionary_for_calibration(
         reference_groups, color_values, performance_type
     )

src/rtichoke/processing/exported_functions.py

@@ -148,6 +148,7 @@ def create_plotly_curve(rtichoke_curve_dict):
         "y": 0,
         "steps": [],
     }
+    sliders_dict["steps"] = []
 
     for k in range(
         len(

src/rtichoke/processing/plotly_helper_functions.py

@@ -5,7 +5,7 @@
 import plotly.graph_objects as go
 import polars as pl
 import math
-from typing import Any, Dict, Union, Sequence
+from typing import Any, Dict, Union, Sequence, cast
 import numpy as np
 from rtichoke.performance_data.performance_data import prepare_performance_data
 from rtichoke.performance_data.performance_data_times import (
@@ -329,8 +329,8 @@ def _create_reference_lines_data(
             # random-guess (y=1 unless all p==0 -> NaN)
             all_zero = (
                 aj_df["p"].len() > 0
-                and float(aj_df["p"].max()) == 0.0
-                and float(aj_df["p"].min()) == 0.0
+                and float(cast(float, aj_df["p"].max())) == 0.0
+                and float(cast(float, aj_df["p"].min())) == 0.0
             )
             rand_y = pl.Series(
                 np.full(len(x_s), np.nan) if all_zero else np.ones(len(x_s)),
@@ -992,7 +992,7 @@ def _check_if_multiple_populations_are_being_validated_times(
         ]
         .max()
     )
-    return max_val is not None and max_val > 1
+    return max_val is not None and float(cast(float, max_val)) > 1
 
 
 def _check_if_multiple_populations_are_being_validated(
@@ -1977,10 +1977,21 @@ def _create_curve_layout(
                 "b": max(80, base_pad.get("b", 0)),
                 **base_pad,
             }
+    xaxis: dict[str, Any] = {"showgrid": False}
+    yaxis: dict[str, Any] = {"showgrid": False}
+
+    if axes_ranges is not None:
+        xaxis["range"] = axes_ranges["xaxis"]
+        yaxis["range"] = axes_ranges["yaxis"]
+
+    if x_label:
+        xaxis["title"] = {"text": x_label}
+    if y_label:
+        yaxis["title"] = {"text": y_label}
 
     curve_layout = {
-        "xaxis": {"showgrid": False},
-        "yaxis": {"showgrid": False},
+        "xaxis": xaxis,
+        "yaxis": yaxis,
         "template": "plotly",
         "plot_bgcolor": "rgba(0, 0, 0, 0)",
         "paper_bgcolor": "rgba(0, 0, 0, 0)",
@@ -2014,15 +2025,6 @@ def _create_curve_layout(
         "modebar": {"remove": list(DEFAULT_MODEBAR_BUTTONS_TO_REMOVE)},
     }
 
-    if axes_ranges is not None:
-        curve_layout["xaxis"]["range"] = axes_ranges["xaxis"]
-        curve_layout["yaxis"]["range"] = axes_ranges["yaxis"]
-
-    if x_label:
-        curve_layout["xaxis"]["title"] = {"text": x_label}
-    if y_label:
-        curve_layout["yaxis"]["title"] = {"text": y_label}
-
     return curve_layout
 
 

src/rtichoke/processing/transforms.py

@@ -67,13 +67,13 @@ def transform_group(group: pl.DataFrame, by: float) -> pl.DataFrame:
 
 def pivot_longer_strata(data: pl.DataFrame) -> pl.DataFrame:
     # Identify id_vars and value_vars
-    id_vars = [col for col in data.columns if not col.startswith("strata_")]
-    value_vars = [col for col in data.columns if col.startswith("strata_")]
+    index_cols = [col for col in data.columns if not col.startswith("strata_")]
+    on_cols = [col for col in data.columns if col.startswith("strata_")]
 
-    # Perform the melt (equivalent to pandas.melt)
-    data_long = data.melt(
-        id_vars=id_vars,
-        value_vars=value_vars,
+    # Perform the unpivot (equivalent to pandas.melt)
+    data_long = data.unpivot(
+        index=index_cols,
+        on=on_cols,
         variable_name="stratified_by",
         value_name="strata",
     )
@@ -257,12 +257,12 @@ def _create_list_data_to_adjust(
         probs_array = np.asarray(probs_dict[reference_group_labels[0]])
 
         if isinstance(reals_dict, dict):
-            reals_array = np.asarray(reals_dict[0])
+            reals_array = np.asarray(reals_dict[reference_group_labels[0]])
         else:
             reals_array = np.asarray(reals_dict)
 
         if isinstance(times_dict, dict):
-            times_array = np.asarray(times_dict[0])
+            times_array = np.asarray(times_dict[reference_group_labels[0]])
         else:
             times_array = np.asarray(times_dict)