-
Notifications
You must be signed in to change notification settings - Fork 0
/
experiments.py
191 lines (147 loc) · 7.37 KB
/
experiments.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
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
import os
import numpy as np
import pandas as pd
import time
import matplotlib.pyplot as plt
from tree.base import DecisionTree
from metrics import *
from sklearn.model_selection import train_test_split
np.random.seed(42)
num_average_time = 100 # Number of times to run each experiment to calculate the average values
# Function to create fake data (take inspiration from usage.py)
def generate_data(N, M, input_type, output_type):
if input_type == "real":
X = pd.DataFrame(np.random.randn(N, M))
elif input_type == "discrete":
X = pd.DataFrame({i: pd.Series(np.random.randint(2, size=N), dtype="category") for i in range(M)})
if output_type == "real":
y = pd.Series(np.random.randn(N))
elif output_type == "discrete":
y = pd.Series(np.random.randint(M, size=N), dtype="category")
return X, y
# Function to calculate average time (and std) taken by fit() and predict() for different N and M for 4 different cases of DTs
def evaluate_runtime(N, M, input_type, output_type, test_size, criterias, num_average_time):
X, y = generate_data(N, M, input_type, output_type)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=test_size)
time_data = {}
for criteria in criterias:
fit_times = []
predict_times = []
for _ in range(num_average_time):
tree = DecisionTree(criterion=criteria, max_depth=5)
start_train = time.time()
tree.fit(X_train, y_train)
fit_times.append(time.time() - start_train)
start_test = time.time()
y_pred = tree.predict(X_test)
predict_times.append(time.time() - start_test)
avg_train_time = np.mean(fit_times)
avg_test_time = np.mean(predict_times)
std_train_time = np.std(fit_times)
std_test_time = np.std(predict_times)
print(f" Criteria: {criteria}")
print(f" Average Training Time: {avg_train_time:.4f} seconds (std: {std_train_time:.4f})")
print(f" Average Prediction Time: {avg_test_time:.4f} seconds (std: {std_test_time:.4f})")
time_data[criteria] = {
"train_time": avg_train_time,
"test_time": avg_test_time,
"std_train_time": std_train_time,
"std_test_time": std_test_time
}
return time_data
def run_n_m(N_values, M_values, input_types, output_types, test_size, criterias, num_average_time):
results = {}
for N in N_values:
for M in M_values:
print(f"\nEvaluating for N={N}, M={M}\n")
results[(N, M)] = {}
for input_type in input_types:
for output_type in output_types:
print(f" Input Type: {input_type}, Output Type: {output_type}")
single_data = evaluate_runtime(N, M, input_type, output_type, test_size, criterias, num_average_time)
results[(N, M)][(input_type, output_type)] = single_data
print()
print("=" * 50)
return results
# Function to plot the results
def plot_time_complexity_separate(results, N_values, M_values, criteria):
# Plot training and prediction time vs N values, keeping M constant
print("Time vs Number of Samples (N)")
plt.figure(figsize=(14, 5*len(M_values)))
plt.suptitle(f"Time vs Number of Samples (N) - Criteria: {criteria}", y=1, fontsize=16)
for i, M in enumerate(M_values):
ax1 = plt.subplot(len(M_values), 2, 2*i + 1)
ax2 = plt.subplot(len(M_values), 2, 2*i + 2)
for input_type in ["discrete", "real"]:
for output_type in ["discrete", "real"]:
train_times = []
prediction_times = []
for N in N_values:
data = results[(N, M)][(input_type, output_type)][criteria]
train_times.append(data["train_time"])
prediction_times.append(data["test_time"])
ax1.plot(N_values, train_times, marker='o', label=f'{input_type.capitalize()}-{output_type.capitalize()}')
ax2.plot(N_values, prediction_times, marker='o', label=f'{input_type.capitalize()}-{output_type.capitalize()}')
ax1.set_xlabel("Number of Samples (N)")
ax1.set_ylabel("Training Time (seconds)")
ax1.set_title(f"Training Time vs Number of Samples (N), M = {M}")
ax1.legend()
ax1.grid(True)
ax2.set_xlabel("Number of Samples (N)")
ax2.set_ylabel("Prediction Time (seconds)")
ax2.set_title(f"Prediction Time vs Number of Samples (N), M = {M}")
ax2.legend()
ax2.grid(True)
plt.savefig(f"./Task-5 Decision Tree Implementation/5.4 Data/5.4_time_vs_N_{criteria}.png", bbox_inches='tight', dpi = 300)
plt.tight_layout()
plt.subplots_adjust(wspace=0.2, hspace=0.25)
plt.show()
print("=" * 50)
# Plot training and prediction time vs M values, keeping N constant
print("Time vs Number of Features (M)")
plt.figure(figsize=(14, 5*len(N_values)))
plt.suptitle(f"Time vs Number of Features (M) - Criteria: {criteria}", y=1, fontsize=16)
for i, N in enumerate(N_values):
ax1 = plt.subplot(len(N_values), 2, 2*i + 1)
ax2 = plt.subplot(len(N_values), 2, 2*i + 2)
for input_type in ["discrete", "real"]:
for output_type in ["discrete", "real"]:
train_times = []
prediction_times = []
for M in M_values:
data = results[(N, M)][(input_type, output_type)][criteria]
train_times.append(data["train_time"])
prediction_times.append(data["test_time"])
ax1.plot(M_values, train_times, marker='o', label=f'{input_type.capitalize()}-{output_type.capitalize()}')
ax2.plot(M_values, prediction_times, marker='o', label=f'{input_type.capitalize()}-{output_type.capitalize()}')
ax1.set_xlabel("Number of Features (M)")
ax1.set_ylabel("Training Time (seconds)")
ax1.set_title(f"Training Time vs Number of Features (M), N = {N}")
ax1.legend()
ax1.grid(True)
ax2.set_xlabel("Number of Features (M)")
ax2.set_ylabel("Prediction Time (seconds)")
ax2.set_title(f"Prediction Time vs Number of Features (M), N = {N}")
ax2.legend()
ax2.grid(True)
plt.savefig(f"./Task-5 Decision Tree Implementation/5.4 Data/5.4_time_vs_M_{criteria}.png", bbox_inches='tight', dpi = 300)
plt.tight_layout()
plt.subplots_adjust(wspace=0.2, hspace=0.25)
plt.show()
# Run the functions, Learn the DTs and Show the results/plots
N_values = [50, 100, 500, 1000, 5000]
M_values = [1, 5, 10, 20, 50, 100]
criterias = ["information_gain", "gini_index"]
input_types = ["real", "discrete"]
output_types = ["real", "discrete"]
test_size = 0.3
results = run_n_m(N_values, M_values, input_types, output_types, test_size, criterias, num_average_time)
# Save the results to a file
import pickle
if not os.path.exists(r'./Task-5 Decision Tree Implementation/5.4 Data'):
os.makedirs(r'./Task-5 Decision Tree Implementation/5.4 Data')
with open(r'./Task-5 Decision Tree Implementation/5.4 Data/5.4_results.pkl', 'wb') as f:
pickle.dump(results, f)
# Plot the results
plot_time_complexity_separate(results, N_values, M_values, criteria="information_gain")
plot_time_complexity_separate(results, N_values, M_values, criteria="gini_index")