12 lines
573 B
Bash
12 lines
573 B
Bash
|
#!/bin/bash
|
||
|
|
kaggle kernels output 'eugenioscionti/scraping-rotten-tomatoes-to-enrich-netflix-dataset'
|
||
|
|
head -n 1 netflix_titles_enriched.csv >> header.csv
|
||
|
|
sed '1d' netflix_titles_enriched.csv | shuf | split -l $[ ($(wc -l netflix_titles_enriched.csv | cut -d" " -f1) - 1) * 60 / 100 ]
|
||
|
|
cat header.csv >> netflix_split_train.csv
|
||
|
|
cat header.csv >> netflix_split_test.csv
|
||
|
|
cat header.csv >> netflix_split_validation.csv
|
||
|
|
cat xaa >> netflix_split_train.csv
|
||
|
|
mv xab split_test_val
|
||
|
|
split -n l/2 split_test_val
|
||
|
|
cat xaa >> netflix_split_test.csv
|
||
|
|
cat xab >> netflix_split_validation.csv
|