-
Notifications
You must be signed in to change notification settings - Fork 3
/
process_results_facebook.py
293 lines (195 loc) · 8.42 KB
/
process_results_facebook.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
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
import pickle
import itertools
import pandas as pd
from functools import reduce
import matplotlib.pyplot as plt
import seaborn as sns
# Loading the files
with open("grid_par_facebook.pickle", "rb") as f:
df = pickle.load(f)
for res in df:
res["mean_reward"] = res["df"]["reward_best"].mean()
max(df, key=lambda x: x["mean_reward"])
# Converting to DataFrames
df = pd.DataFrame(df)
df["rewards"] = df.apply(lambda x: [x["df"]["reward_best"]], axis=1)
df_test = df[["sigma", "c", "epsilon", "mean_reward", "rewards"]]
df_test = df_test.sort_values("mean_reward", ascending=False)
# %% Making a heatmap for sigma/c, new map for every epsilon
fig = plt.figure(figsize=(20, 20))
subplots = fig.subplots(3, 2)
subplots_cont = itertools.chain([x for i in subplots for x in i])
for epsilon in sorted(df_test["epsilon"].unique()):
axis = next(subplots_cont)
axis.set_title(f"Epsilon = {epsilon}")
df_heat = df_test[df_test["epsilon"] == epsilon][["sigma", "c", "mean_reward"]]
sns.heatmap(df_heat.pivot("sigma", "c", "mean_reward"), ax=axis, vmin=5, vmax=9)
# %% Processing the num_repeats results
with open("num_reps_facebook.pickle", "rb") as f:
df_num_repeats = pickle.load(f)
df_num_repeats
for res in df_num_repeats:
res["mean_reward"] = res["df"]["reward_best"].mean()
df_num_repeats = pd.DataFrame(df_num_repeats)
sns.lineplot(df_num_repeats["num_repeats"], df_num_repeats["mean_reward"])
# %% Processing num_repeats_reward results
with open("num_reps_reward_facebook.pickle", "rb") as f:
df_num_repeats_rew = pickle.load(f)
for res in df_num_repeats_rew:
res["mean_reward"] = res["df"]["reward_best"].mean()
df_num_repeats_rew = pd.DataFrame(df_num_repeats_rew)
sns.lineplot(
df_num_repeats_rew["num_repeats_reward"], df_num_repeats_rew["mean_reward"]
)
# %% Processing the RSB results
with open("grid_rsb.pickle", "rb") as f:
df_rsb = pickle.load(f)
for res in df_rsb:
res["mean_reward"] = res["df"]["reward"].mean()
max(df_rsb, key=lambda x: x["mean_reward"])
df_rsb = pd.DataFrame(df_rsb)
df_rsb["rewards"] = df_rsb.apply(lambda x: [x["df"]["reward"]], axis=1)
df_rsb = df_rsb.sort_values("mean_reward", ascending=False)
df_heat = df_rsb[["gamma", "c", "mean_reward"]]
sns.heatmap(df_heat.pivot("gamma", "c", "mean_reward")) # , vmin=5, vmax=11)
# %% Processing num_repeats_reward results for RSB
with open("num_repeats_rsb_facebook.pickle", "rb") as f:
df_num_repeats_rsb = pickle.load(f)
for res in df_num_repeats_rsb:
res["mean_reward"] = res["df"]["reward"].mean()
df_num_repeats_rsb = pd.DataFrame(df_num_repeats_rsb)
df_num_repeats_rsb_plot = df_num_repeats_rsb[
["num_repeats_expect", "mean_reward"]
].melt("num_repeats_expect", var_name="cols", value_name="values")
sns.lineplot(x="num_repeats_expect", y="values", data=df_num_repeats_rsb_plot)
# %% Processing the comparison for 5 seeds (FB dataset)
with open("comparison_facebook_5seeds.pickle", "rb") as f:
df_comp = pickle.load(f)
df_comp["tim_t"]
df_comp["rsb_persist"]
df_comp["rsb_nppersist"]
df_comp["timlinucb"]
for key, df in df_comp.items():
df.columns = df.columns.map(lambda x: str(x) + "_" + key if x != "time" else x)
df_comp = reduce(lambda left, right: pd.merge(left, right, on="time"), df_comp.values())
df_comp_plot = df_comp[
[
"time",
"reward_tim_t",
"reward_rsb_persist",
"reward_rsb_nppersist",
"reward_best_timlinucb",
]
].melt("time", var_name="Algorithm", value_name="Reward")
sns.lineplot(x="time", y="Reward", hue="Algorithm", data=df_comp_plot)
# %% Processing the comprison for 20 seeds, 100 time steps
with open("comparison_facebook_20seeds_100t.pickle", "rb") as f:
df_comp_100t = pickle.load(f)
df_comp_100t
df_comp_100t
for df in df_comp_100t:
algo_name = df["algo_name"][0]
df.columns = df.columns.map(
lambda x: str(x) + "_" + algo_name if x != "time" else x
)
df_comp_100t = reduce(
lambda left, right: pd.merge(left, right, on="time"), df_comp_100t
)
df_comp_100t_plot = df_comp_100t[
[
"time",
"reward_tim_t",
"reward_rsb_persist",
"reward_rsb_nopersist",
"reward_best_timlinucb",
]
].melt("time", var_name="Algorithm", value_name="Reward")
sns.lineplot(x="time", y="Reward", hue="Algorithm", data=df_comp_100t_plot)
# %% Checking persistent vs non-persistent parameters (20 time steps)
with open("comparison_facebook_20seeds.pickle", "rb") as f:
df_comp = pickle.load(f)
with open("tlu_20days_persist_facebook.pickle", "rb") as f:
df_tlu_persist = pickle.load(f)
df_tlu_nopersist = df_comp[3][["reward_best", "time", "s_best"]]
df_tlu_nopersist.rename(
columns={"reward_best": "reward_nopersist", "s_best": "s_nopersist"}, inplace=True
)
df_tlu_persist = df_tlu_persist[["reward_best", "time", "s_best"]]
df_tlu_persist.rename(
columns={"reward_best": "reward_persist", "s_best": "s_persist"}, inplace=True
)
df_persist_nopersist = pd.merge(df_tlu_nopersist, df_tlu_persist, on="time")
df_persist_nopersist_plot = df_persist_nopersist[
["time", "reward_persist", "reward_nopersist",]
].melt("time", var_name="Algorithm", value_name="Reward")
sns.lineplot(x="time", y="Reward", hue="Algorithm", data=df_persist_nopersist_plot)
# %% Checking persistent vs non-persistent parameters (100 time steps)
with open("comparison_facebook_20seeds_100t.pickle", "rb") as f:
df_comp = pickle.load(f)
with open("tlu_100days_persist_facebook.pickle", "rb") as f:
df_tlu_persist = pickle.load(f)
df_tlu_nopersist = df_comp[3][["reward_best", "time", "s_best"]]
df_tlu_nopersist.rename(
columns={"reward_best": "reward_nopersist", "s_best": "s_nopersist"}, inplace=True
)
df_tlu_persist = df_tlu_persist[["reward_best", "time", "s_best"]]
df_tlu_persist.rename(
columns={"reward_best": "reward_persist", "s_best": "s_persist"}, inplace=True
)
df_persist_nopersist = pd.merge(df_tlu_nopersist, df_tlu_persist, on="time")
df_persist_nopersist_plot = df_persist_nopersist[
["time", "reward_persist", "reward_nopersist",]
].melt("time", var_name="Algorithm", value_name="Reward")
sns.lineplot(x="time", y="Reward", hue="Algorithm", data=df_persist_nopersist_plot)
# --------------------------------------------------------------------------------------
# %% -------------------------------- DIGG DATASET -------------------------------------
# --------------------------------------------------------------------------------------
with open("grid_rsb_digg.pickle", "rb") as f:
df_rsb = pickle.load(f)
for res in df_rsb:
res["mean_reward"] = res["df"]["reward"].mean()
max(df_rsb, key=lambda x: x["mean_reward"])
df_rsb = pd.DataFrame(df_rsb)
df_rsb["rewards"] = df_rsb.apply(lambda x: [x["df"]["reward"]], axis=1)
df_rsb = df_rsb.sort_values("mean_reward", ascending=False)
df_heat = df_rsb[["gamma", "c", "mean_reward"]]
sns.heatmap(df_heat.pivot("gamma", "c", "mean_reward")) # , vmin=5, vmax=11)
# %% TLU for digg
with open("grid_par_digg.pickle", "rb") as f:
df = pickle.load(f)
for res in df:
res["mean_reward"] = res["df"]["reward_best"].mean()
max(df, key=lambda x: x["mean_reward"])
# Converting to DataFrames
df = pd.DataFrame(df)
df["rewards"] = df.apply(lambda x: [x["df"]["reward_best"]], axis=1)
df_test = df[["sigma", "c", "epsilon", "mean_reward", "rewards"]]
df_test = df_test.sort_values("mean_reward", ascending=False)
fig = plt.figure(figsize=(20, 20))
subplots = fig.subplots(3, 2)
subplots_cont = itertools.chain([x for i in subplots for x in i])
for epsilon in sorted(df_test["epsilon"].unique()):
axis = next(subplots_cont)
axis.set_title(f"Epsilon = {epsilon}")
df_heat = df_test[df_test["epsilon"] == epsilon][["sigma", "c", "mean_reward"]]
sns.heatmap(df_heat.pivot("sigma", "c", "mean_reward"), ax=axis)
# %% Processing the comparison for 20 seeds/100t (Digg dataset)
with open("comparison_digg_20seeds_100t.pickle", "rb") as f:
df_comp = pickle.load(f)
df_comp["tim_t"]
df_comp["rsb_persist"]
df_comp["rsb_nppersist"]
df_comp["timlinucb"]
for key, df in df_comp.items():
df.columns = df.columns.map(lambda x: str(x) + "_" + key if x != "time" else x)
df_comp = reduce(lambda left, right: pd.merge(left, right, on="time"), df_comp.values())
df_comp_plot = df_comp[
[
"time",
"reward_tim_t",
"reward_rsb_persist",
"reward_rsb_nppersist",
"reward_best_timlinucb",
]
].melt("time", var_name="Algorithm", value_name="Reward")
sns.lineplot(x="time", y="Reward", hue="Algorithm", data=df_comp_plot)