Add 'dataset-stats.py'

This commit is contained in:
Tomasz Koszarek 2023-09-29 05:13:07 +02:00
parent a501c1af9c
commit 6767fc1ab4

18
dataset-stats.py Normal file
View File

@ -0,0 +1,18 @@
import pandas
data_train = pandas.read_csv('results/train_data.csv',
engine='python',
encoding='ISO-8859-1',
sep=',')
data_dev = pandas.read_csv('results/dev_data.csv',
engine='python',
encoding='ISO-8859-1',
sep=',')
data_test = pandas.read_csv('results/test_data.csv',
engine='python',
encoding='ISO-8859-1',
sep=',')
data_train.describe(include='all').to_csv('train_data_stats.csv', index=True)
data_dev.describe(include='all').to_csv('dev_data_stats.csv', index=True)
data_test.describe(include='all').to_csv('test_data_stats.csv', index=True)