Zaktualizuj 'download.sh'

This commit is contained in:
Szymon Parafiński 2022-03-27 22:41:36 +02:00
parent 520f8e797c
commit 6ba48875e9

View File

@ -1,15 +1,17 @@
dataset_operation() {
tail -n +2 dataset.csv | shuf > dataset.csv.s
head -n $CUTOFF dataset.csv.s > ./dataset.csv.shuf
len1=$(cat ./imdb-dataset.csv.shuf | wc -l)
tail -n +2 all_games.csv | shuf > all_games.csv.s
head -n $CUTOFF all_games.csv.s > ./all_games.csv.shuf
len1=$(cat ./all_games.csv.shuf | wc -l)
len2=$(($len1/10))
len3=$(($len2*2))
len4=$(($len3+1))
head -n $len2 dataset.csv.shuf > dataset_test.csv
head -n $len3 dataset.csv.shuf | tail -n $len2 > dataset_dev.csv.
tail -n +$len4 dataset.csv.shuf > dataset_train.csv
rm dataset.csv.shuf
wc -l dataset.csv.*
head -n $len2 all_games.csv.shuf > all_games.csv.test
head -n $len3 all_games.csv.shuf | tail -n $len2 > all_games.csv.dev
tail -n +$len4 all_games.csv.shuf > all_games.csv.train
rm all_games.csv.shuf
wc -l all_games.csv.*
}
kaggle datasets download -d deepcontractor/top-video-games-19952021-metacritic
kaggle datasets download -d deepcontractor/top-video-games-19952021-metacritic
unzip top-video-games-19952021-metacritic.zip
dataset_operation