ium_z444510/create-dataset.sh
2023-04-21 00:02:37 +02:00

15 lines
647 B
Bash

#!/bin/bash
export KAGGLE_USERNAME=$KAGGLE_USERNAME
export KAGGLE_KEY=$KAGGLE_KEY
kaggle datasets download -d thedevastator/airbnb-prices-in-european-cities
unzip airbnb-prices-in-european-cities.zip -d data
rm airbnb-prices-in-european-cities.zip
head -n $CUTOFF data/barcelona_weekends.csv | shuf > data/barcelona_weekends.shuf.csv
head -n $CUTOFF/4 data/barcelona_weekends.shuf.csv > data/barcelona_weekends.test.csv
head -n $CUTOFF/2 data/barcelona_weekends.shuf.csv | tail -n 25 > data/barcelona_weekends.dev.csv
tail -n $CUTOFF/2+1 data/barcelona_weekends.shuf.csv > data/barcelona_weekends.train.csv
rm data/barcelona_weekends.shuf.csv