added plotter for results

plotter/reward_plotter.py

@@ -0,0 +1,39 @@
+import numpy as np
+import matplotlib.pyplot as plt
+import os
+
+def plot_csv(paths, x_axis, y_axis):
+    for path_ in paths:
+        data = np.genfromtxt(path_, delimiter=',', skip_header=1, dtype=float)
+
+        mean = np.mean(data, axis=1)
+        std = np.std(data, axis=1)
+
+        x = np.linspace(0, mean.shape[0], mean.shape[0])
+
+        # Extract the first part of the filename and use it as a label
+        label = os.path.basename(path_).split('-')[0:3]
+        label = f"{label[0]} {float(label[1].replace('_','.'))}, nrbfs = {int(label[2])}"
+
+        plt.plot(x, mean, label=label)
+        plt.fill_between(
+            x,
+            mean - 1.96 * std,
+            mean + 1.96 * std,
+            alpha=0.5
+        )
+    plt.xlabel(x_axis)
+    plt.xlim([0, mean.shape[0]])
+    plt.ylabel(y_axis)
+    plt.grid(True)
+    plt.legend(loc="best")
+    plt.show()
+
+
+if __name__ == '__main__':
+
+    filenames = ['random-1_0-5-1685552722_2243946.csv']
+    home_dir = os.path.expanduser('~')
+    file_path = os.path.join(home_dir, 'Documents/IntRLResults')
+    paths = [os.path.join(file_path, filename) for filename in filenames]
+    plot_csv(paths, 'Episodes', 'Reward')

runner/BOGymRunner.py

@@ -5,7 +5,8 @@ import numpy as np
 import matplotlib.pyplot as plt
 
 # from ToyTask.MountainCarGym import Continuous_MountainCarEnv
-from ToyTask.Pendulum import PendulumEnv
+# from ToyTask.Pendulum import PendulumEnv
+from ToyTask.Cartpole import CartPoleEnv
 
 import warnings
 from sklearn.exceptions import ConvergenceWarning
@@ -13,12 +14,12 @@ from sklearn.exceptions import ConvergenceWarning
 warnings.filterwarnings("ignore", category=ConvergenceWarning)
 
 # BO parameters
-env = PendulumEnv()
+env = CartPoleEnv()
 nr_steps = 100
 acquisition_fun = 'ei'
-iteration_steps = 100
+iteration_steps = 50
 
-nr_runs = 100
+nr_runs = 10
 
 # storage arrays
 finished_store = np.zeros((1, nr_runs))
@@ -82,7 +83,7 @@ def main():
     global finished_store, best_policy, reward_store
     bo = BayesianOptimization(env, nr_steps, acq=acquisition_fun)
     for i in range(nr_runs):
-        print('Iteration:', str(i))
+        print('Runs:', str(i))
         bo.env_seed = int(np.random.randint(1, 2147483647, 1)[0])
         bo.initialize()
         for j in range(iteration_steps):