#!/bin/bash

kaggle datasets download -d lava18/google-play-store-apps
unzip -o google-play-store-apps.zip
sed -i '1d' googleplaystore.csv
shuf googleplaystore.csv > apps_shuf_.csv
head -n $CUTOFF apps_shuf_.csv > apps_shuf.csv

total=$(wc -l apps_shuf.csv | awk '{print $1}')
test_set=$(( (total*60+50) / 100 )) 
train_val_set=$(( (total*20+50) / 100 )) 

#cp apps_shuf.csv apps_shuf_copy.csv

head -n $test_set apps_shuf.csv > apps_train.csv
lines="1,$test_set"
sed -i "$lines"'d' apps_shuf.csv
head -n $train_val_set apps_shuf.csv > apps_test.csv
lines="1,$train_val_set"
sed -i "$lines"'d' apps_shuf.csv
head -n $train_val_set apps_shuf.csv > apps_validate.csv