ium_464863/Jenkinsfile

54 lines
1.5 KiB
Plaintext
Raw Normal View History

2024-03-20 14:23:42 +01:00
pipeline {
2024-03-29 16:59:06 +01:00
agent any
parameters {
password (
name: 'KAGGLE_USERNAME',
defaultValue: '',
description: 'Kaggle username'
2024-03-20 17:30:31 +01:00
)
password (
name: 'KAGGLE_KEY',
defaultValue: '',
description: 'Kaggle API key'
2024-03-20 17:30:31 +01:00
)
2024-03-20 17:33:00 +01:00
string (
name: 'CUTOFF',
defaultValue: '500',
description: 'Get only the first CUTOFF rows of the dataset'
)
}
2024-03-20 14:23:42 +01:00
stages {
stage('Clone repository') {
steps {
checkout scm
}
}
stage('Download dataset') {
2024-03-20 14:23:42 +01:00
steps {
2024-03-29 16:59:06 +01:00
withEnv(["KAGGLE_USERNAME=${params.KAGGLE_USERNAME}", "KAGGLE_KEY=${params.KAGGLE_KEY}"]) {
sh "kaggle datasets download -d uciml/breast-cancer-wisconsin-data"
sh "unzip -o breast-cancer-wisconsin-data.zip"
2024-03-29 17:08:38 +01:00
sh "mkdir -p datasets"
sh "mv data.csv datasets/data.csv"
}
2024-03-20 14:23:42 +01:00
}
}
2024-03-29 16:59:06 +01:00
stage('Preprocess data') {
agent {
dockerfile {
filename 'Dockerfile'
}
}
steps {
2024-03-29 17:02:48 +01:00
sh "chmod +x ./download_dataset.py"
2024-03-29 16:59:06 +01:00
sh "python3 ./download_dataset.py ${params.CUTOFF}"
archiveArtifacts artifacts: './datasets/data.csv,./datasets/train.csv,./datasets/dev.csv,./datasets/test.csv', onlyIfSuccessful: true
}
}
2024-03-20 14:23:42 +01:00
}
}