2023-04-20 23:03:49 +02:00
|
|
|
#!/bin/bash
|
2023-04-20 22:27:59 +02:00
|
|
|
|
2023-04-20 23:15:39 +02:00
|
|
|
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
|
2023-04-21 00:04:47 +02:00
|
|
|
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
|
2023-04-20 23:15:39 +02:00
|
|
|
rm data/barcelona_weekends.shuf.csv
|
