100 lines
4.0 KiB
Python
100 lines
4.0 KiB
Python
import numpy as np
|
|
import matplotlib.pyplot as plt
|
|
import os
|
|
|
|
|
|
def plot_csv_with_titles(paths_dict, x_axis, y_axis, subplot_titles, y_limits=(0, 200)):
|
|
# Adjustments for dynamic subplot creation based on the dictionary size
|
|
num_subplots = len(paths_dict)
|
|
fig, axs = plt.subplots(1, num_subplots, figsize=(4*num_subplots, 4))
|
|
|
|
# If only one subplot, axs is not an array, so we need to convert it
|
|
if num_subplots == 1:
|
|
axs = [axs]
|
|
|
|
for idx, (_, file_list) in enumerate(paths_dict.items()):
|
|
for path_, color in file_list:
|
|
data = np.genfromtxt(path_, delimiter=',', skip_header=0, dtype=float)
|
|
|
|
if data.shape[0]> data.shape[1]:
|
|
mean = np.mean(data, axis=1)
|
|
std = np.std(data, axis=1)
|
|
else:
|
|
mean = np.mean(data, axis=0)
|
|
std = np.std(data, axis=0)
|
|
|
|
x = np.linspace(0, mean.shape[0], mean.shape[0])
|
|
|
|
axs[idx].plot(x, mean, color=color)
|
|
axs[idx].fill_between(
|
|
x,
|
|
mean - 1.96 * std,
|
|
mean + 1.96 * std,
|
|
color=color,
|
|
alpha=0.5
|
|
)
|
|
axs[idx].set_title(subplot_titles[idx])
|
|
axs[idx].set_xlabel(x_axis)
|
|
axs[idx].set_xlim([0, mean.shape[0]])
|
|
axs[idx].grid(True)
|
|
axs[idx].set_ylim(y_limits)
|
|
|
|
# Only label the y-axis for the leftmost plot
|
|
if idx == 0:
|
|
axs[idx].set_ylabel(y_axis)
|
|
else:
|
|
axs[idx].set_ylabel('')
|
|
|
|
plt.tight_layout()
|
|
plt.show()
|
|
|
|
|
|
if __name__ == '__main__':
|
|
|
|
home_dir = os.path.expanduser('~')
|
|
file_path = os.path.join(home_dir, 'Documents/IntRLResults/CP-Results')
|
|
|
|
filenames = [
|
|
'cp-e150r10-bf15-base/cp-ei-random-1_0-15-1690282051_2959082.csv',
|
|
'cp-e150r10-bf15-noshaping/cp-pei-random-0_95-15-1690276946_1944933.csv',
|
|
'cp-e150r10-bf15-noshaping/cp-pei-regular-25_0-15-1690290021_6843266.csv',
|
|
'cp-e150r10-bf15-noshaping/cp-pei-improvement-0_1-15-1690292664_0382216.csv',
|
|
'cp-e150r10-bf15-shaping/cp-ei-random-0_95-15-1690451164_0115042.csv',
|
|
'cp-e150r10-bf15-shaping/cp-ei-regular-25_0-15-1690456185_1115792.csv',
|
|
'cp-e150r10-bf15-shaping/cp-ei-improvement-0_1-15-1690465143_0114875.csv',
|
|
'cp-e150r10-bf15-shaping/cp-pei-random-0_95-15-1690467921_7118568.csv',
|
|
'cp-e150r10-bf15-shaping/cp-pei-regular-25_0-15-1690470012_0117908.csv',
|
|
'cp-e150r10-bf15-shaping/cp-pei-improvement-0_1-15-1690472449_4115295.csv',
|
|
]
|
|
filepaths = [os.path.join(file_path, filename) for filename in filenames]
|
|
|
|
# Demonstrating the adjusted function with subplot titles
|
|
titles = ["Preference", "Shaping", "Combination", "Regular"]
|
|
|
|
data_dict_colored = {
|
|
'subplot1': [(filepaths[0], 'C0'), (filepaths[1], 'C1'), (filepaths[2], 'C2'), (filepaths[3], 'C3')],
|
|
'subplot2': [(filepaths[0], 'C0'), (filepaths[4], 'C1'), (filepaths[5], 'C2'), (filepaths[6], 'C3')],
|
|
'subplot3': [(filepaths[0], 'C0'), (filepaths[7], 'C1'), (filepaths[8], 'C2'), (filepaths[9], 'C3')],
|
|
'subplot4': [(filepaths[0], 'C0'), (filepaths[2], 'C6'), (filepaths[5], 'C8'), (filepaths[8], 'C5')],
|
|
}
|
|
|
|
plot_csv_with_titles(data_dict_colored, 'Episodes', 'Reward', titles)
|
|
|
|
file_path_reacher = os.path.join(home_dir, 'Documents/IntRLResults/RE-Results')
|
|
filenames_reacher = ['base_line/re-ei-random-1_0-5-1694370994_0363934.csv',
|
|
'shaping/re-ei-random-1_0-10-1694359559_616903.csv',
|
|
'shaping/re-ei-regular-10_0-5-1694371946_5364418.csv'
|
|
]
|
|
|
|
filepaths_reacher = [os.path.join(file_path_reacher, filename) for filename in filenames_reacher]
|
|
|
|
titles_reacher = ["Shaping"]
|
|
|
|
data_dict_reacher = {
|
|
'subplot1': [(filepaths_reacher[0], 'C0'), (filepaths_reacher[1], 'C1'), (filepaths_reacher[2], 'C2')]
|
|
}
|
|
|
|
plot_csv_with_titles(data_dict_reacher, 'Episodes', 'Reward', titles_reacher, y_limits=(-150, 50))
|
|
|
|
|