2022-03-28 00:36:00 +02:00
|
|
|
kaggle datasets download -d csafrit2/steel-industry-energy-consumption --force
|
2022-03-28 00:34:49 +02:00
|
|
|
unzip -o steel-industry-energy-consumption.zip
|
2022-03-27 23:51:53 +02:00
|
|
|
|
2022-03-28 00:52:40 +02:00
|
|
|
head -n $CUTOFF Steel_industry_data.csv | tail -n +2 | cut -d, -f8 --complement | shuf > steel_industry_data_shuffled.csv
|
2022-03-27 23:51:53 +02:00
|
|
|
number_of_lines=$(wc -l steel_industry_data_shuffled.csv | awk '{print $1}')
|
|
|
|
|
test_len=$((number_of_lines/10))
|
|
|
|
|
dev_len=$((number_of_lines/10))
|
|
|
|
|
head -n $test_len steel_industry_data_shuffled.csv > steel_industry_data_test.csv
|
|
|
|
|
tail -n +$((test_len+1)) steel_industry_data_shuffled.csv | head -n $dev_len > steel_industry_data_dev.csv
|
|
|
|
|
tail -n +$((test_len+dev_len+1)) steel_industry_data_shuffled.csv > steel_industry_data_train.csv
|
|
|
|
|
rm steel_industry_data_shuffled.csv
|
