add catboost different feature length

figure_31_how_much_data_is_needed.py

@@ -8,7 +8,7 @@
 PATH_PER = '/Users/Timon/Library/CloudStorage/OneDrive-Charité-UniversitätsmedizinBerlin/Shared Documents - ICN Data World/General/Data/UCSF_OLARU/out_per'
 PATH_FIGURE = '/Users/Timon/Library/CloudStorage/OneDrive-Charité-UniversitätsmedizinBerlin/Shared Documents - ICN Data World/General/Data/UCSF_OLARU/figures_ucsf'
 
-for label in ["tremor", "DK", "bk"]:
+for label in ["bk", "tremor", "DK",]:
     files = [f for f in os.listdir(PATH_PER) if "LOHO" in f and"_min.pkl" in f and "_"+label+"_" in f]
     df_ = []
     for f in files:
@@ -37,7 +37,8 @@
 
     df = pd.concat(df_, axis=0)
     # clip the balanced accuracy to 0.5 and 1
-    df["per"] = np.clip(df["per"], 0.5, 1)
+    if label != "bk":
+        df["per"] = np.clip(df["per"], 0.5, 1)
 
     df.groupby("dur")["per"].mean()
 
@@ -52,14 +53,21 @@
         plt.plot(durations / 60, df_sub["per"], color="gray", alpha=0.2)
         sub_per.append(df_sub["per"].values)
     plt.xlabel("Duration [h]")
-    plt.ylabel("Balanced accuracy")
+    if label == "bk":
+        plt.ylabel("Correlation coefficient")
+    else:
+        plt.ylabel("Balanced accuracy")
     # plot the mean accuracy for each duration
     plt.plot(durations / 60, np.array(sub_per).mean(axis=0), marker="o", linestyle="-", color="black")
+    # write the mean accuracy on top of the line
+    for i, dur in enumerate(durations):
+        plt.text(durations[i] / 60, np.array(sub_per).mean(axis=0)[i], f"{np.round(np.array(sub_per).mean(axis=0)[i], 2)}", ha="center", va="bottom")
+
     plt.xscale('log')
-    plt.title("LOHO PKG DK CLASS CV different training duration")
+    plt.title(f"LOHO PKG BK CV different training duration")
     plt.tight_layout()
     plt.savefig(os.path.join(PATH_FIGURE, f"LOHO_different_training_duration_sub_{label}.pdf"))
-    plt.show(block=True)
+    #plt.show(block=True)
 
 plt.figure(figsize=(10, 5))
 sns.boxplot(data=df, x="dur", y="per", palette="viridis", showfliers=False, showmeans=True)

figure_32_CatBoost_multiple_timeslots.py

@@ -0,0 +1,121 @@
+import os
+import pandas as pd
+import numpy as np
+import seaborn as sns
+from matplotlib import pyplot as plt
+import pickle
+
+PATH_PER = '/Users/Timon/Library/CloudStorage/OneDrive-Charité-UniversitätsmedizinBerlin/Shared Documents - ICN Data World/General/Data/UCSF_OLARU/out_per'
+PATH_FIGURE = '/Users/Timon/Library/CloudStorage/OneDrive-Charité-UniversitätsmedizinBerlin/Shared Documents - ICN Data World/General/Data/UCSF_OLARU/figures_ucsf'
+
+df_mean_all = []
+df_ = []
+for label in ["bk", "tremor", "dk",]:
+    files = [f for f in os.listdir(PATH_PER) if "cont_data_CatBOOST" in f and".pkl" in f and "_"+label+"_" in f]
+
+    for f in files:
+        with open(os.path.join(PATH_PER, f), "rb") as f_:
+            d_out = pickle.load(f_)
+
+            l = []
+            if "CLASS_True" in f:
+                CLASS = True
+                per_label = "ba"
+            else:
+                CLASS = False
+                per_label = "corr_coeff"
+            if "_dk_" in f:
+                label_name = "pkg_dk"
+            elif "_tremor_" in f:
+                label_name = "pkg_tremor"
+            elif "_bk_" in f:
+                label_name = "pkg_bk"
+
+            str_ = f[f.find("CatBOOST"):]
+            dur = int(str_.split("_")[1])
+
+            #for loc_ in d_out[CLASS][label_name].keys():
+            loc_ = "ecog_stn"
+            for sub_test in d_out[CLASS][label_name][loc_].keys():
+                l.append({
+                    "sub": sub_test,
+                    "pkg_label": label_name,
+                    "CLASS": CLASS,
+                    "dur" : dur,
+                    "per": d_out[CLASS][label_name][loc_][sub_test][per_label]
+                })
+            df = pd.DataFrame(l)
+            df_.append(df)
+
+        # df = pd.concat(df_, axis=0)
+        # get mean grouped by duration
+        # df_mean = df.groupby("dur")["per"].mean()
+        # df_mean = df_mean.reset_index()
+        # df_mean["pkg_label"] = label
+        # df_mean["CLASS"] = CLASS
+        # df_mean_all.append(df_mean)
+
+df = pd.concat(df_, axis=0)
+df.groupby(["pkg_label", "CLASS", "dur",])["per"].mean()
+
+plt.figure(figsize=(5, 10))
+idx_ = 0
+for label in df["pkg_label"].unique():
+    for CLASS in df["CLASS"].unique():
+        idx_ += 1
+        plt.subplot(3, 2, idx_)
+        df_plt = df.query(f"CLASS == {CLASS} and pkg_label == '{label}' and dur < 100").copy().reset_index()
+        sns.boxplot(x="dur", y="per", data=df_plt, showmeans=False, palette="viridis", showfliers=False)
+        sns.swarmplot(x="dur", y="per", data=df_plt, color=".25", palette="viridis")
+        # write the mean on top of the boxplot
+        df_mean = df_plt.groupby("dur")["per"].mean()
+        for i, mean in enumerate(df_mean):
+            plt.text(i, mean, f"{mean:.2f}", ha="center", va="center", color="white")
+        plt.xlabel("Duration [min]")
+        if CLASS:
+            plt.ylabel("Balanced accuracy")
+        else:
+            plt.ylabel("Correlation coefficient")
+        plt.title(f"{label} - {CLASS}")
+
+#plt.plot(df_plt.groupby("dur")["per"].mean().values, label=f"{label} - {CLASS}")
+#plt.xticks(range(len(df_plt["dur"].unique())), np.sort(df_plt["dur"].unique()))
+plt.tight_layout()
+plt.show(block=True)
+
+df_plt = df.query(f"CLASS == True and pkg_label == '{label}'")
+spec = dict(x="dur", y="per", data=df_plt)
+sns.stripplot(**spec, size=4, color=".7")
+sns.pointplot(**spec, errorbar=None, linestyle="none", marker="_", markersize=30)
+plt.show(block=True)
+
+
+
+plt.figure()
+idx_ = 0
+for label in df["pkg_label"].unique():
+    for CLASS in df["CLASS"].unique():
+        df_plt = df.query(f"CLASS == {CLASS} and pkg_label == '{label}'")
+        idx_ += 1
+        plt.subplot(3, 2, idx_)
+        for sub in df_plt["sub"].unique():
+            df_sub = df_plt.query(f"sub == '{sub}'")
+            df_sub = df_sub.sort_values("dur")
+            plt.plot(df_sub["dur"], df_sub["per"], color="gray", alpha=0.2)
+        # plot the mean
+        df_mean = df_plt.groupby("dur")["per"].mean()
+        plt.plot(df_mean, label=f"{label} - {CLASS}")
+
+        plt.xlabel("Duration [min]")
+        # log scale
+        #plt.xscale("log")
+        plt.title(f"{label} - {CLASS}")
+        if CLASS == True:
+            plt.ylabel("Balanced accuracy")
+            plt.ylim(0.5, 1)
+        else:
+            plt.ylabel("Correlation coefficient")
+        plt.xlim(0, 100)
+
+plt.tight_layout()
+plt.show(block=True)

how_much_data_is_best_for_CatBoost.py

@@ -147,9 +147,9 @@ def compute_duration(dur, CLASS, label_name):
     for label_idx, label in enumerate(label_names):
         for CLASS_Idx, CLASS in enumerate(CLASSES):
             for comb_dura_idx, dur in enumerate(comb_durations):
-                if cnt == idx_:
-                    compute_duration(dur, CLASS, label)
-                cnt = cnt + 1
+                #if cnt == idx_:
+                compute_duration(dur, CLASS, label)
+                #cnt = cnt + 1
 
 
     #for duration in comb_durations: