forked from TeaPearce/Counter-Strike_Behavioural_Cloning
-
Notifications
You must be signed in to change notification settings - Fork 0
/
tools_dataset_stats.py
201 lines (168 loc) · 7.17 KB
/
tools_dataset_stats.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
192
193
194
195
196
197
198
199
import numpy as np
import matplotlib.pyplot as plt
import matplotlib as mpl
plt.rcParams['pdf.fonttype'] = 42
plt.rcParams['ps.fonttype'] = 42
plt.rcParams.update({'font.size': 16})
save_path = '/Users/timpearce/Google Drive/Google Drive all/05. misc geeky/03_csgo_bot/01_writeup/images/aaaishots/'
# figsize_in = (5,2.5) # x, y width
is_save=False
file_name_stub = 'dm_july2021_'
# folder_name = 'G:/2021/csgo_bot_train_july2021/'
# folder_name = '/Volumes/My Passport/2021/csgo_bot_train_july2021/'
folder_name = '/Volumes/My Passport/2021/csgo_bot_train_july2021/dataset_metadata/'
n_filer_per_chunk=100
info_array = []
weap_arr=[]
weap_type_arr=[]
for file_chunk in range(0,3):
# for file_chunk in range(0,55):
print('file_chunk',file_chunk,end='\r')
dict_chunk = np.load(folder_name+'currvarsv2_'+file_name_stub+str(file_chunk*n_filer_per_chunk+1)+'_to_'+str((file_chunk+1)*n_filer_per_chunk)+'.npy',allow_pickle=True)
dict_chunk = dict_chunk.item()
# file_num1000_frame_0
# file_num1000_frame_999
# use this to make sure go through in order
prev_mouse = (0,0)
for file_i in range(file_chunk*n_filer_per_chunk+1,(file_chunk+1)*n_filer_per_chunk+1):
for frame_i in range(0,1000):
key = 'file_num' + str(file_i) +'_frame_' + str(frame_i)
dict_key = dict_chunk[key]
mousex = dict_key[1][1]
mousey = dict_key[1][2]
pos1 = dict_key[0]['localpos1']
pos2 = dict_key[0]['localpos2']
pos3 = dict_key[0]['localpos3']
kill_flag = dict_key[2][0]
death_flag = dict_key[2][1]
if 'gsi_weap_active' in dict_key[0].keys():
weap_arr.append(dict_key[0]['gsi_weap_active']['name'])
if 'taser' not in dict_key[0]['gsi_weap_active']['name']:
weap_type_arr.append(dict_key[0]['gsi_weap_active']['type'])
else:
weap_type_arr.append('taser')
else:
weap_arr.append('none found')
weap_type_arr.append('none found')
if 'ak47' in weap_arr[-1]:
ak_flag=1
else:
ak_flag=0
# if (mousex,mousey) == prev_mouse and mousex != 0:
if mousex < prev_mouse[0]*1.1 and mousex > prev_mouse[0]*0.9 and mousex != 0:
same_mouse_flag = 1
else:
same_mouse_flag = 0
info_array.append([pos1,pos2,pos3,mousex,mousey,kill_flag,death_flag,ak_flag,same_mouse_flag])
prev_mouse = (mousex,mousey)
info_array = np.array(info_array)
kill_arr = info_array[info_array[:,5]==1]
death_arr = info_array[info_array[:,6]==1]
ak_arr = info_array[info_array[:,7]==1]
not_ak_arr = info_array[info_array[:,7]!=1]
same_mouse_arr = info_array[info_array[:,8]==1]
# info_array = ak_arr # could overwrie
# info_array = not_ak_arr
print('total frames',info_array.shape[0])
print('total kills',info_array[:,5].sum())
print('total deaths',info_array[:,6].sum())
print('total ak frames',info_array[:,7].sum())
print('total ak kills',ak_arr[:,5].sum())
print('total ak deaths',ak_arr[:,6].sum())
print('mean ak kills',ak_arr[:,5].mean())
print('mean ak deaths',ak_arr[:,6].mean())
weap_count_dict = {}
for w in weap_arr:
if w in weap_count_dict.keys():
weap_count_dict[w] +=1
else:
weap_count_dict[w] = 1
type_count_dict = {}
for w in weap_type_arr:
if w in type_count_dict.keys():
type_count_dict[w] +=1
else:
type_count_dict[w] = 1
knife_total=0
for w in weap_count_dict.keys():
if 'knife' in w:
knife_total+=weap_count_dict[w]
weap_count_dict[w]=0
weap_count_dict['weapon_knives'] = knife_total
total=0
for w in weap_count_dict.keys():
if 'm4a1' in w:
total+=weap_count_dict[w]
weap_count_dict[w]=0
weap_count_dict['weapon_m4a1'] = total
weap_count_arr = []
weap_count_name = []
for w in weap_count_dict.keys():
if w == 'none found':
pass
else:
weap_count_name.append(w[7:])
weap_count_arr.append(weap_count_dict[w])
weap_count_arr = np.array(weap_count_arr)
weap_count_name = np.array(weap_count_name)
ids_order = np.argsort(weap_count_arr)
weap_count_arr = weap_count_arr[ids_order]
weap_count_name = weap_count_name[ids_order]
weap_count_arr = weap_count_arr/weap_count_arr.sum()
fig, ax = plt.subplots(1,1,figsize=(8,8))
# ax.scatter(info_array[:100000,0],info_array[:100000,1],alpha=0.02,s=10,lw=0.,color='magenta')
# ax.scatter(info_array[:,0],info_array[:,1],alpha=0.005,s=5,lw=0.,color='magenta')
ax.scatter(info_array[:2000000,0],info_array[:2000000,1],alpha=0.005,s=4,lw=0.,color='magenta')
# ax.scatter(death_arr[:,0],death_arr[:,1],alpha=0.1,s=40,lw=0.,color='red')
# ax.scatter(kill_arr[:,0],kill_arr[:,1],alpha=0.1,s=40,lw=0.,color='blue')
ax.set_xlabel('Player coords x')
ax.set_ylabel('Player coords y')
ax.set_xticks([])
ax.set_yticks([])
fig.show()
name = save_path + 'dataset_trajs'
if is_save:
fig.savefig(name+'.png', format='png', dpi=500, bbox_inches='tight')
fig, ax = plt.subplots(1,1)
h = ax.hist2d(info_array[:100000,0],info_array[:100000,1],bins=30,cmap='jet',norm=mpl.colors.LogNorm())
fig.show()
fig.colorbar(h[3], ax=ax)
fig, ax = plt.subplots(1,1,figsize=(7,3))
ax.hist(np.clip(info_array[:,3],-380,380),alpha=1,bins=200, edgecolor=None,color='deepskyblue',density=True,histtype='bar',zorder=2)
ax.hist(np.clip(info_array[:,3],-380,380),alpha=1,bins=200, edgecolor=None,color='k',density=True,histtype='step',lw=1,zorder=3)
for x in [ -300.0, -200.0, -100.0, -60.0, -30.0, -20.0, -10.0, -4.0, -2.0, -0.0, 2.0, 4.0, 10.0, 20.0, 30.0, 60.0, 100.0, 200.0, 300.0]:
ax.axvline(x,zorder=1,color='k',lw=0.5,ls='--')
ax.set_xlabel('Mouse x')
ax.set_ylabel('Density')
ax.set_xlim((-350,350))
ax.set_yticks([])
fig.show()
name = save_path + 'dataset_mousex'
if is_save:
fig.savefig(name+'.pdf', format='pdf', dpi=1000, bbox_inches='tight')
fig, ax = plt.subplots(1,1,figsize=(7,3))
ax.hist(np.clip(info_array[:,4],-380,380),alpha=1,bins=200, edgecolor=None,color='deepskyblue',density=True,histtype='bar',zorder=2)
ax.hist(np.clip(info_array[:,4],-380,380),alpha=1,bins=200, edgecolor=None,color='k',density=True,histtype='step',lw=1,zorder=3)
for x in [-100.0, -50.0, -20.0, -10.0, -4.0, -2.0, -0.0, 2.0, 4.0, 10.0, 20.0, 50.0, 100.0]:
ax.axvline(x,zorder=1,color='k',lw=0.5,ls='--')
ax.set_xlabel('Mouse y')
ax.set_ylabel('Density')
ax.set_xlim((-350,350))
ax.set_yticks([])
fig.show()
name = save_path + 'dataset_mousey'
if is_save:
fig.savefig(name+'.pdf', format='pdf', dpi=1000, bbox_inches='tight')
nshow=25
weap_count_arr[-nshow] = weap_count_arr[:-nshow].sum()
weap_count_name[-nshow] = 'other'
fig, ax = plt.subplots(1,1,figsize=(8,8))
ax.grid(which='both',color='k', linestyle='--', linewidth=1,alpha=0.2,axis='x',markevery=0.05,zorder=1)
ax.barh(np.arange(len(weap_count_arr[-nshow:])), weap_count_arr[-nshow:], align='center', edgecolor='black',color='deepskyblue',zorder=2)
ax.set_yticks(np.arange(len(weap_count_arr[-nshow:])))
ax.set_yticklabels(weap_count_name[-nshow:])
ax.set_xlabel('Proportion of time equipped')
fig.show()
name = save_path + 'dataset_equip'
if is_save:
fig.savefig(name+'.pdf', format='pdf', dpi=1000, bbox_inches='tight')