diff --git a/download_new.sh b/download_new.sh new file mode 100644 index 0000000..b043203 --- /dev/null +++ b/download_new.sh @@ -0,0 +1,17 @@ +dataset_operation() { + 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 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 +unzip top-video-games-19952021-metacritic.zip +dataset_operation \ No newline at end of file