2021-03-28 20:07:35 +02:00
|
|
|
#!/bin/bash
|
|
|
|
|
2021-03-28 20:24:44 +02:00
|
|
|
curl -LO https://git.wmi.amu.edu.pl/s430705/ium_430705/raw/branch/master/imdb_movies.csv
|
2021-03-28 21:19:20 +02:00
|
|
|
shuf ./imdb_movies.csv > ./imdb_movies2.csv | tail -n +$((${CUTOFF})) > ./imdb_movies2.csv
|
2021-03-28 21:05:57 +02:00
|
|
|
wc -l ./imdb_movies.csv
|
|
|
|
head -n 4600 ./imdb_movies2.csv > ./test.csv
|
|
|
|
head -n 4600 ./imdb_movies2.csv | tail -n 4600 > ./dev.csv
|
|
|
|
tail -n 13800 ./imdb_movies2.csv > ./train.csv
|
|
|
|
wc -l ./*.csv
|