-
Notifications
You must be signed in to change notification settings - Fork 1
/
plotting.py
126 lines (98 loc) · 4.71 KB
/
plotting.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
"""
Define plotting functions here.
"""
# STD
import os
# EXT
import numpy as np
import matplotlib.pyplot as plt
import matplotlib
# PROJECT
from analyze import test_difference
if os.environ.get('DISPLAY', '') == '':
print('no display found. Using non-interactive Agg backend')
matplotlib.use('Agg')
def smooth(x, N):
cumsum = np.cumsum(np.insert(x, 0, 0))
return (cumsum[N:] - cumsum[:-N]) / float(N)
def plot_exp_performance(exps, path):
for exp, env_name in exps:
plt.figure()
for episode_duration, exp_name in exp:
plt.plot(smooth(episode_duration, 10), label=exp_name)
plt.title('Episode durations per episode in ' + env_name)
plt.legend()
plt.savefig(f"{path}/{env_name}.png")
def plot_exps_with_intervals(q_data: np.array, dq_data: np.array, file_name, title=None, significant_values=None,
true_q: float=None, true_dq: float=None, smooth_curves=False):
"""
Plot scores with intervals. Expects a K x D matrix with K trials with D data points each.
"""
assert q_data.shape[0] > 1 and dq_data.shape[0] > 1, "At least two trials per model are necessary to create this plot!"
def get_curves(data):
median = np.median(data, axis=0)
# Average of the two most extreme values to get upper / lower bounds
upper = np.mean(np.sort(data, axis=0)[-2:, :], axis=0)
lower = np.mean(np.sort(data, axis=0)[:2, :], axis=0)
if smooth_curves:
return smooth(lower, 10), smooth(median, 10), smooth(upper, 10)
else:
return lower, median, upper
# Get the median, lower and upper curves
q_lower, q_median, q_upper = get_curves(q_data)
dq_lower, dq_median, dq_upper = get_curves(dq_data)
# Plot everything
x = np.arange(0, q_lower.shape[0])
plt.plot(q_median, label="DQN", color="firebrick")
plt.fill_between(x, q_upper, q_median, facecolor="lightcoral", alpha=0.6)
plt.fill_between(x, q_median, q_lower, facecolor="lightcoral", alpha=0.6)
plt.plot(dq_median, label="Double DQN", color="navy")
plt.fill_between(x, dq_upper, dq_median, facecolor="lightsteelblue", alpha=0.6)
plt.fill_between(x, dq_median, dq_lower, facecolor="lightsteelblue", alpha=0.6)
# If the true values are given, plot them as a straight line
if true_q is not None:
plt.axhline(true_q, label="True DQN value", color="firebrick", linestyle='dashed', alpha=0.8)
if true_dq is not None:
plt.axhline(true_dq, label="True Double DQN value", color="lightsteelblue", linestyle='dashed', alpha=0.8)
# Emphasize significant values if given
if significant_values is not None:
num_points = q_data.shape[0] * q_data.shape[1]
lowest_data_point = np.min([np.min(q_data.reshape(num_points)), np.min(dq_data.reshape(num_points))])
lowest_data_point = int(lowest_data_point) - 5 if lowest_data_point != 0 else -10
plt.scatter(
significant_values, np.ones(significant_values.shape) * lowest_data_point,
marker="|", color="black"
)
# Emphasize significant values if given
if significant_values is not None:
plt.scatter(significant_values, np.zeros(significant_values.shape), marker="", color="black", alpha=0.7)
if title is not None:
plt.title(title)
plt.legend(fontsize=8)
plt.tight_layout()
plt.savefig(file_name)
plt.close()
def plot_data_for_env(env_name, q_data, dq_data, image_path):
q_values, q_rewards, q_durations, true_q = q_data["values"], q_data["rewards"], q_data["durations"], q_data["true"]
dq_values, dq_rewards, dq_durations, true_dq = dq_data["values"], dq_data["rewards"], dq_data["durations"], dq_data["true"]
# Do significance-testing
print(env_name)
print("Q-values")
_, significant_scores = test_difference(q_values, dq_values)
print("Rewards")
_, significant_rewards = test_difference(q_rewards, dq_rewards)
print("Durations")
_, significant_durations = test_difference(q_durations, dq_durations)
print("")
plot_exps_with_intervals(
q_values, dq_values, title=f"{env_name} Q-Values", file_name=f"{image_path}/qvalues_{env_name.lower()}.png",
smooth_curves=False, true_q=true_q, true_dq=true_dq, significant_values=significant_scores
)
plot_exps_with_intervals(
q_rewards, dq_rewards, title=f"{env_name} Rewards", file_name=f"{image_path}/rewards_{env_name.lower()}.png",
smooth_curves=False, significant_values=significant_rewards
)
plot_exps_with_intervals(
q_durations, dq_durations, title=f"{env_name} Durations", file_name=f"{image_path}/durations_{env_name.lower()}.png",
smooth_curves=False, significant_values=significant_durations
)