init
Browse files
stats.py
CHANGED
@@ -101,8 +101,13 @@ if __name__ == '__main__':
|
|
101 |
e_dist_full.append(e_dist)
|
102 |
data_size_full.append(data_size)
|
103 |
config.append([min_e_freq, max_p_freq])
|
104 |
-
|
105 |
-
|
|
|
|
|
|
|
|
|
|
|
106 |
|
107 |
# check statistics
|
108 |
print("- Data Size")
|
@@ -119,20 +124,15 @@ if __name__ == '__main__':
|
|
119 |
# plot predicate distribution
|
120 |
df_p = pd.DataFrame([dict(enumerate(sorted(p.values(), reverse=True))) for p in p_dist_full]).T
|
121 |
df_p.columns = [f"min entity: {mef}, max predicate: {mpf}" for mef, mpf in candidates]
|
122 |
-
plt.figure()
|
123 |
fig, axes = plt.subplots(2, 2, constrained_layout=True)
|
124 |
fig.suptitle('Predicate Distribution over Different Configurations')
|
125 |
for (x, y), mpf in zip([(0, 0), (0, 1), (1, 0), (1, 1)], [100, 50, 25, 10]):
|
126 |
-
# fig = plt.figure()
|
127 |
_df = df_p[[f"min entity: {mef}, max predicate: {mpf}" for mef in [1, 2, 3, 4]]]
|
128 |
_df.columns = [f"min entity: {mef}" for mef in [1, 2, 3, 4]]
|
129 |
ax = sns.lineplot(ax=axes[x, y], data=_df, linewidth=1)
|
130 |
if mpf != 100:
|
131 |
ax.legend_.remove()
|
132 |
axes[x, y].set_title(f'max predicate: {mpf}')
|
133 |
-
# ax.set(xlabel='unique predicates sorted by frequency', ylabel='number of triples', title='Predicate Distribution (max predicate: {mpf})')
|
134 |
-
# ax.get_figure().savefig(f"data/stats.predicate_distribution.max_predicate_{mpf}.png", bbox_inches='tight')
|
135 |
-
# ax.get_figure().clf()
|
136 |
fig.supxlabel('unique predicates sorted by frequency')
|
137 |
fig.supylabel('number of triples')
|
138 |
fig.savefig("data/stats.predicate_distribution.png", bbox_inches='tight')
|
|
|
101 |
e_dist_full.append(e_dist)
|
102 |
data_size_full.append(data_size)
|
103 |
config.append([min_e_freq, max_p_freq])
|
104 |
+
# save data
|
105 |
+
for s in ['train', 'validation', 'test']:
|
106 |
+
new_data_s = [i for i in new_data if i['split'] == s]
|
107 |
+
for i in new_data_s:
|
108 |
+
i.pop('split')
|
109 |
+
with open(f"data/t_rex.clean.min_entity_{min_e_freq}_max_predicate_{max_p_freq}.{s}.jsonl", 'w') as f:
|
110 |
+
f.write('\n'.join([json.dumps(i) for i in new_data_s]))
|
111 |
|
112 |
# check statistics
|
113 |
print("- Data Size")
|
|
|
124 |
# plot predicate distribution
|
125 |
df_p = pd.DataFrame([dict(enumerate(sorted(p.values(), reverse=True))) for p in p_dist_full]).T
|
126 |
df_p.columns = [f"min entity: {mef}, max predicate: {mpf}" for mef, mpf in candidates]
|
|
|
127 |
fig, axes = plt.subplots(2, 2, constrained_layout=True)
|
128 |
fig.suptitle('Predicate Distribution over Different Configurations')
|
129 |
for (x, y), mpf in zip([(0, 0), (0, 1), (1, 0), (1, 1)], [100, 50, 25, 10]):
|
|
|
130 |
_df = df_p[[f"min entity: {mef}, max predicate: {mpf}" for mef in [1, 2, 3, 4]]]
|
131 |
_df.columns = [f"min entity: {mef}" for mef in [1, 2, 3, 4]]
|
132 |
ax = sns.lineplot(ax=axes[x, y], data=_df, linewidth=1)
|
133 |
if mpf != 100:
|
134 |
ax.legend_.remove()
|
135 |
axes[x, y].set_title(f'max predicate: {mpf}')
|
|
|
|
|
|
|
136 |
fig.supxlabel('unique predicates sorted by frequency')
|
137 |
fig.supylabel('number of triples')
|
138 |
fig.savefig("data/stats.predicate_distribution.png", bbox_inches='tight')
|