solve estimation error

Liphos · Jan 25, 2023 · a534f48 · a534f48
1 parent 346693f
commit a534f48
Show file tree

Hide file tree

Showing 2 changed files with 2 additions and 19 deletions.
diff --git a/core/main.py b/core/main.py
@@ -314,8 +314,8 @@ def compute_bins(pred: np.ndarray, energy: np.ndarray) -> Tuple[np.ndarray, np.n
     pred_mean, pred_std, true_mean = [], [], []
     for e_bin in enumerate(bins):
         indicies = np.where(ind_bins == e_bin[0])[0]
-        pred_mean.append(np.mean(pred[:, indicies] - energy[indicies]))
-        pred_std.append(np.sqrt(np.mean(np.std(pred[:, indicies], axis=0)**2)))
+        pred_mean.append(np.mean((pred[:, indicies] - energy[indicies])/energy[indicies]))
+        pred_std.append(np.sqrt(np.mean(np.std((pred[:, indicies] - energy[indicies])/energy[indicies], axis=0)**2)))
         true_mean.append(np.mean(energy[indicies]))
 
     pred_mean = np.array(pred_mean)

diff --git a/core/plot_utils.py b/core/plot_utils.py
@@ -83,29 +83,12 @@ def plot_bins_results(train_values: Tuple[np.ndarray, np.ndarray, np.ndarray],
 
     #Plot the results on the bins
     plt.clf()
-    #TODO: Error instead of dividing of taking the mean of (E_pr - e_th)/e_th we did mean(E_pr - E_th)/mean(E_th)
     plt.errorbar(true_train_mean, pred_train_mean, yerr=pred_train_std, fmt="o", label="Train")
     plt.plot(true_train_mean, [0 for _ in range(len(true_train_mean))], "k")
     plt.errorbar(true_test_mean, pred_test_mean, yerr=pred_test_std, fmt="o", label="Val")
     plt.plot(true_test_mean, [0 for _ in range(len(true_test_mean))], "k")
     plt.title("Results")
     plt.xlabel("ground truth energy (EeV)")
-    plt.ylabel("$E_{pr} - E_{th} (EeV)$")
-    plt.xlim(0, 4.1)
-    plt.legend()
-    if fig_dir is not None:
-        plt.savefig(fig_dir + "/" + "all")
-
-    #Plot the residue
-    plt.figure()
-    plt.errorbar(true_train_mean, pred_train_mean/true_train_mean,
-                    yerr=pred_train_std/true_train_mean, fmt="o", label="Train")
-    plt.plot(true_train_mean, [0 for _ in range(len(true_train_mean))], "k")
-    plt.errorbar(true_test_mean, pred_test_mean/true_test_mean,
-                    yerr=pred_test_std/true_test_mean, fmt="o", label="Val")
-    plt.plot(true_test_mean, [0 for _ in range(len(true_test_mean))], "k")
-    plt.title("Results")
-    plt.xlabel("ground truth energy (EeV)")
     plt.ylabel(r"$\frac{E_{pr} - E_{th}}{E_{th}} $")
     plt.xlim(0, 4.1)
     plt.legend()