mixUCB with Reward Rational Choice

compare_params_NOA.py

@@ -0,0 +1,153 @@
+
+import pandas as pd
+import matplotlib.pyplot as plt 
+import seaborn as sns
+import numpy as np
+import re
+
+import ast
+
+
+
+def plot_compare_linePlot(df, metrics):
+    feedback_types = df["feedback_type"].unique()
+    n = len(feedback_types)
+
+    for metric in metrics:
+        # Create figure with n rows: one per feedback type
+        fig, axs = plt.subplots(n, 2, figsize=(14, 4 * n), sharex=False)
+        fig.suptitle(f"{metric.replace('_', ' ').title()} — by Feedback Type", fontsize=16)
+
+        for i, feedback_type in enumerate(feedback_types):
+            # LINE PLOT
+            ax_line = axs[i, 0]
+            for (source, expert_type), group in df[df["feedback_type"] == feedback_type].groupby(["source", "expert_type"]):
+                label = f"{source}: {expert_type}"
+                ax_line.plot(group["delta"], group[metric], marker="o", label=label)
+            ax_line.set_title(f"{feedback_type} — Line Plot")
+            ax_line.set_ylabel(metric.replace('_', ' ').title())
+            ax_line.set_xlabel("Delta")
+            ax_line.grid(True)
+            if i == 0:
+                ax_line.legend(title="Expert Type", bbox_to_anchor=(1.05, 1), loc="upper left", fontsize=9)
+
+            # HEATMAP
+            ax_heat = axs[i, 1]
+            pivot = df[df["feedback_type"] == feedback_type].pivot_table(
+                index="expert_type", columns="delta", values=metric, aggfunc="mean"
+            )
+            sns.heatmap(pivot, annot=True, fmt=".2f", cmap="viridis", ax=ax_heat)
+            ax_heat.set_title(f"{feedback_type} — Heatmap")
+            ax_heat.set_xlabel("Delta")
+            ax_heat.set_ylabel("Expert Type")
+
+        plt.tight_layout(rect=[0, 0, 1, 0.96])
+        plt.show()
+
+
+
+def plot_timeseries(df, metrics):
+    """
+    df: DataFrame with columns [delta, expert_type, rewards_per_time, queries_per_time]
+    deltas: list of delta values to plot
+    metrics: which timeseries to plot
+    """
+    for delta in set(df["delta"]):
+
+        for feedback_type in set(df["feedback_type"]):
+
+            fig, axs = plt.subplots(1, len(metrics), figsize=(6*len(metrics), 4))
+            if len(metrics) == 1:
+                axs = [axs]
+            fig.suptitle(f"Δ = {delta}, Feedback_type = {feedback_type}", fontsize=14)
+
+            for j, metric in enumerate(metrics):
+                ax = axs[j]
+
+                # for expert_type, group in df.groupby("expert_type"):
+                #     series = np.vstack(group[metric].values)  # stack runs
+                #     mean = series.mean(axis=0)
+                #     std = series.std(axis=0)
+                #     ax.plot(mean, label=expert_type)
+                #     ax.fill_between(np.arange(len(mean)), mean-std, mean+std, alpha=0.2)
+
+                for expert_type, group in df.groupby("expert_type"):
+                    subset = group.loc[(group["delta"] == delta) & (group["feedback_type"] == feedback_type)]
+                    ax.plot(subset[metric].iloc[0], label=expert_type)
+
+                ax.set_title(metric.replace("_", " ").title())
+                ax.set_xlabel("time steps")
+                ax.legend()
+
+            plt.tight_layout(rect=[0, 0, 1, 0.95])
+            plt.show()
+
+
+
+def ensure_array(x):
+    if isinstance(x, str):
+        # remove "np.float64(" and the closing ")"
+        cleaned = x.replace("np.float64(", "").replace(")", "").replace(" ","").replace("[","").replace("]","")
+        # now extract numbers (handles floats and ints)
+        x_list = cleaned.split(",")
+        return np.array([float(n) for n in x_list], dtype=float)
+    elif isinstance(x, list):
+        return np.array(x, dtype=float)
+    elif isinstance(x, np.ndarray):
+        return x.astype(float)
+    else:
+        raise ValueError(f"Unexpected type {type(x)}")
+
+
+
+
+########################################################################################
+
+# INIT:
+
+
+# files = {
+#     "OG": "MixUCB_compare_21Aug2025.xlsx",
+#     "EXP4": "MixUCB_compare_EXP4_21Aug2025.xlsx",
+#     "MetaChoice": "MixUCB_compare_MetaChoice_21Aug2025.xlsx"
+# }
+files = {
+    "OG": "MixUCB_compare.xlsx",
+    "EXP4": "MixUCB_compare_EXP4.xlsx",
+    "MetaChoice": "MixUCB_compare_MetaChoice.xlsx"
+}
+
+metrics = ["cum_auto_rewards_per_time", "cum_queries_per_time"]
+
+
+########################################################################################
+
+
+
+# CODE:
+
+
+# df = pd.read_excel("MixUCB_compare_Jul11.xlsx")
+
+# Load and tag each DataFrame
+dfs = []
+for label, filepath in files.items():
+    df = pd.read_excel(filepath)
+    df["source"] = label  # Add source column
+    dfs.append(df)
+
+# Concatenate all into one DataFrame
+df_all = pd.concat(dfs, ignore_index=True)
+
+#conversion step (because in excel list is saved as str):
+for col in metrics:
+    if col in df_all.columns:
+        df_all[col] = df_all[col].apply(ensure_array)
+
+
+
+# plot_compare_heatmap(df_all)
+
+# plot_compare_linePlot(df_all, metrics=["total_reward", "query_num", "total_auto_reward"])
+
+plot_timeseries(df_all, metrics)