Terjemahan disediakan oleh mesin penerjemah. Jika konten terjemahan yang diberikan bertentangan dengan versi bahasa Inggris aslinya, utamakan versi bahasa Inggris.
# Menggunakan APIs untuk mengukur dan mengelola kualitas data
Topik ini menjelaskan cara menggunakan APIs untuk mengukur dan mengelola kualitas data.
**Contents**
+ [Prasyarat](#using-apis-prerequisites)
+ [Bekerja dengan rekomendasi Kualitas Data AWS Glue](#using-apis-recommendations)
+ [Bekerja dengan Aturan Kualitas Data AWS Glue](#using-apis-rulesets)
+ [Bekerja dengan AWS Glue Data Quality berjalan](#using-apis-runs)
+ [Bekerja dengan hasil Kualitas Data AWS Glue](#using-apis-results)
## Prasyarat
+ Pastikan versi boto3 Anda mutakhir sehingga menyertakan API Kualitas Data AWS Glue terbaru.
+ Pastikan versi AWS CLI Anda mutakhir, sehingga menyertakan CLI terbaru.
Jika Anda menggunakan pekerjaan AWS Glue untuk menjalankan ini APIs, Anda dapat menggunakan opsi berikut untuk memperbarui pustaka boto3 ke versi terbaru:
```
—additional-python-modules boto3==
```
## Bekerja dengan rekomendasi Kualitas Data AWS Glue
**Untuk memulai rekomendasi AWS Glue Data Quality, jalankan:**
```
class GlueWrapper:
"""Encapsulates AWS Glue actions."""
def __init__(self, glue_client):
"""
:param glue_client: A Boto3 AWS Glue client.
"""
self.glue_client = glue_client
def start_data_quality_rule_recommendation_run(self, database_name, table_name, role_arn):
"""
Starts a recommendation run that is used to generate rules when you don't know what rules to write. AWS Glue Data Quality
analyzes the data and comes up with recommendations for a potential ruleset. You can then triage the ruleset
and modify the generated ruleset to your liking.
:param database_name: The name of the AWS Glue database which contains the dataset.
:param table_name: The name of the AWS Glue table against which we want a recommendation
:param role_arn: The Amazon Resource Name (ARN) of an AWS Identity and Access Management (IAM) role that grants permission to let AWS Glue access the resources it needs.
"""
try:
response = self.client.start_data_quality_rule_recommendation_run(
DataSource={
'GlueTable': {
'DatabaseName': database_name,
'TableName': table_name
}
},
Role=role_arn
)
except ClientError as err:
logger.error(
"Couldn't start data quality recommendation run %s. Here's why: %s: %s", name,
err.response['Error']['Code'], err.response['Error']['Message'])
raise
else:
return response['RunId']
```
Untuk menjalankan rekomendasi, Anda dapat menggunakan `pushDownPredicates` atau `catalogPartitionPredicates` untuk meningkatkan kinerja dan menjalankan rekomendasi hanya pada partisi tertentu dari sumber katalog Anda.
```
client.start_data_quality_rule_recommendation_run(
DataSource={
'GlueTable': {
'DatabaseName': database_name,
'TableName': table_name,
'AdditionalOptions': {
'pushDownPredicate': "year=2022"
}
}
},
Role=role_arn,
NumberOfWorkers=2,
CreatedRulesetName=''
)
```
**Untuk mendapatkan hasil rekomendasi AWS Glue Data Quality, jalankan:**
```
class GlueWrapper:
"""Encapsulates AWS Glue actions."""
def __init__(self, glue_client):
"""
:param glue_client: A Boto3 AWS Glue client.
"""
self.glue_client = glue_client
def get_data_quality_rule_recommendation_run(self, run_id):
"""
Gets the specified recommendation run that was used to generate rules.
:param run_id: The id of the data quality recommendation run
"""
try:
response = self.client.get_data_quality_rule_recommendation_run(RunId=run_id)
except ClientError as err:
logger.error(
"Couldn't get data quality recommendation run %. Here's why: %s: %s", run_id,
err.response['Error']['Code'], err.response['Error']['Message'])
raise
else:
return response
```
Dari objek respon di atas, Anda dapat mengekstrak RuleSet yang direkomendasikan oleh run, untuk digunakan dalam langkah selanjutnya:
```
print(response['RecommendedRuleset'])
Rules = [
RowCount between 2000 and 8000,
IsComplete "col1",
IsComplete "col2",
StandardDeviation "col3" between 58138330.8 and 64258155.09,
ColumnValues "col4" between 1000042965 and 1214474826,
IsComplete "col5"
]
```
**Untuk mendapatkan daftar semua rekomendasi Anda berjalan yang dapat difilter dan dicantumkan:**
```
response = client.list_data_quality_rule_recommendation_runs(
Filter={
'DataSource': {
'GlueTable': {
'DatabaseName': '',
'TableName': ''
}
}
)
```
**Untuk membatalkan tugas rekomendasi Kualitas Data AWS Glue yang ada:**
```
response = client.cancel_data_quality_rule_recommendation_run(
RunId='dqrun-d4b6b01957fdd79e59866365bf9cb0e40fxxxxxxx'
)
```
## Bekerja dengan Aturan Kualitas Data AWS Glue
**Untuk membuat aturan Kualitas Data AWS Glue:**
```
response = client.create_data_quality_ruleset(
Name='',
Ruleset='Rules = [IsComplete "col1", IsPrimaryKey "col2", RowCount between 2000 and 8000]',
TargetTable={
'TableName': '',
'DatabaseName': ''
}
)
```
**Untuk mendapatkan aturan kualitas data:**
```
response = client.get_data_quality_ruleset(
Name=''
)
print(response)
```
Anda dapat menggunakan API ini untuk kemudian mengekstrak set aturan:
```
print(response['Ruleset'])
```
**Untuk membuat daftar semua aturan kualitas data untuk tabel:**
```
response = client.list_data_quality_rulesets()
```
Anda dapat menggunakan kondisi filter dalam API untuk memfilter semua aturan yang dilampirkan ke database atau tabel tertentu:
```
response = client.list_data_quality_rulesets(
Filter={
'TargetTable': {
'TableName': '',
'DatabaseName': ''
}
},
)
```
**Untuk memperbarui kumpulan aturan kualitas data:**
```
class GlueWrapper:
"""Encapsulates AWS Glue actions."""
def __init__(self, glue_client):
"""
:param glue_client: A Boto3 AWS Glue client.
"""
self.glue_client = glue_client
def update_data_quality_ruleset(self, ruleset_name, ruleset_string):
"""
Update an AWS Glue Data Quality Ruleset
:param ruleset_name: The name of the AWS Glue Data Quality ruleset to update
:param ruleset_string: The DQDL ruleset string to update the ruleset with
"""
try:
response = self.client.update_data_quality_ruleset(
Name=ruleset_name,
Ruleset=ruleset_string
)
except ClientError as err:
logger.error(
"Couldn't update the AWS Glue Data Quality ruleset. Here's why: %s: %s",
err.response['Error']['Code'], err.response['Error']['Message'])
raise
else:
return response
```
**Untuk menghapus kumpulan aturan kualitas data:**
```
class GlueWrapper:
"""Encapsulates AWS Glue actions."""
def __init__(self, glue_client):
"""
:param glue_client: A Boto3 AWS Glue client.
"""
self.glue_client = glue_client
def delete_data_quality_ruleset(self, ruleset_name):
"""
Delete a AWS Glue Data Quality Ruleset
:param ruleset_name: The name of the AWS Glue Data Quality ruleset to delete
"""
try:
response = self.client.delete_data_quality_ruleset(
Name=ruleset_name
)
except ClientError as err:
logger.error(
"Couldn't delete the AWS Glue Data Quality ruleset. Here's why: %s: %s",
err.response['Error']['Code'], err.response['Error']['Message'])
raise
else:
return response
```
## Bekerja dengan AWS Glue Data Quality berjalan
**Untuk memulai menjalankan AWS Glue Data Quality:**
```
class GlueWrapper:
"""Encapsulates AWS Glue actions."""
def __init__(self, glue_client):
"""
:param glue_client: A Boto3 AWS Glue client.
"""
self.glue_client = glue_client
def start_data_quality_ruleset_evaluation_run(self, database_name, table_name, role_name, ruleset_list):
"""
Start an AWS Glue Data Quality evaluation run
:param database_name: The name of the AWS Glue database which contains the dataset.
:param table_name: The name of the AWS Glue table against which we want to evaluate.
:param role_arn: The Amazon Resource Name (ARN) of an AWS Identity and Access Management (IAM) role that grants permission to let AWS Glue access the resources it needs.
:param ruleset_list: The list of AWS Glue Data Quality ruleset names to evaluate.
"""
try:
response = client.start_data_quality_ruleset_evaluation_run(
DataSource={
'GlueTable': {
'DatabaseName': database_name,
'TableName': table_name
}
},
Role=role_name,
RulesetNames=ruleset_list
)
except ClientError as err:
logger.error(
"Couldn't start the AWS Glue Data Quality Run. Here's why: %s: %s",
err.response['Error']['Code'], err.response['Error']['Message'])
raise
else:
return response['RunId']
```
Ingat bahwa Anda dapat meneruskan `catalogPartitionPredicate` parameter `pushDownPredicate` atau untuk memastikan kualitas data Anda berjalan hanya menargetkan sekumpulan partisi tertentu dalam tabel katalog Anda. Contoh:
```
response = client.start_data_quality_ruleset_evaluation_run(
DataSource={
'GlueTable': {
'DatabaseName': '',
'TableName': '',
'AdditionalOptions': {
'pushDownPredicate': 'year=2023'
}
}
},
Role='',
NumberOfWorkers=5,
Timeout=123,
AdditionalRunOptions={
'CloudWatchMetricsEnabled': False
},
RulesetNames=[
'',
]
)
```
Anda juga dapat mengonfigurasi bagaimana aturan komposit dalam kumpulan aturan Anda dievaluasi, baik di tingkat ROW atau COLUMN. Untuk informasi lebih lanjut tentang cara kerja aturan komposit, silakan lihat [Cara kerja aturan komposit](dqdl.md#dqdl-syntax-composite-rules) dalam dokumentasi.
Contoh tentang cara mengatur metode evaluasi aturan komposit dalam permintaan Anda:
```
response = client.start_data_quality_ruleset_evaluation_run(
DataSource={
'GlueTable': {
'DatabaseName': '',
'TableName': '',
'AdditionalOptions': {
'pushDownPredicate': 'year=2023'
}
}
},
Role='',
NumberOfWorkers=5,
Timeout=123,
AdditionalRunOptions={
'CompositeRuleEvaluationMethod':ROW
},
RulesetNames=[
'',
]
)
```
**Untuk mendapatkan informasi tentang AWS Glue Data Quality run:**
```
class GlueWrapper:
"""Encapsulates AWS Glue actions."""
def __init__(self, glue_client):
"""
:param glue_client: A Boto3 AWS Glue client.
"""
self.glue_client = glue_client
def get_data_quality_ruleset_evaluation_run(self, run_id):
"""
Get details about an AWS Glue Data Quality Run
:param run_id: The AWS Glue Data Quality run ID to look up
"""
try:
response = self.client.get_data_quality_ruleset_evaluation_run(
RunId=run_id
)
except ClientError as err:
logger.error(
"Couldn't look up the AWS Glue Data Quality run ID. Here's why: %s: %s",
err.response['Error']['Code'], err.response['Error']['Message'])
raise
else:
return response
```
**Untuk mendapatkan hasil dari AWS Glue Data Quality menjalankan:**
Untuk menjalankan Kualitas Data AWS Glue tertentu, Anda dapat mengekstrak hasil evaluasi run menggunakan metode berikut:
```
response = client.get_data_quality_ruleset_evaluation_run(
RunId='d4b6b01957fdd79e59866365bf9cb0e40fxxxxxxx'
)
resultID = response['ResultIds'][0]
response = client.get_data_quality_result(
ResultId=resultID
)
print(response['RuleResults'])
```
**Untuk mencantumkan semua AWS Glue Data Quality Anda berjalan:**
```
class GlueWrapper:
"""Encapsulates AWS Glue actions."""
def __init__(self, glue_client):
"""
:param glue_client: A Boto3 AWS Glue client.
"""
self.glue_client = glue_client
def list_data_quality_ruleset_evaluation_runs(self, database_name, table_name):
"""
Lists all the AWS Glue Data Quality runs against a given table
:param database_name: The name of the database where the data quality runs
:param table_name: The name of the table against which the data quality runs were created
"""
try:
response = self.client.list_data_quality_ruleset_evaluation_runs(
Filter={
'DataSource': {
'GlueTable': {
'DatabaseName': database_name,
'TableName': table_name
}
}
}
)
except ClientError as err:
logger.error(
"Couldn't list the AWS Glue Quality runs. Here's why: %s: %s",
err.response['Error']['Code'], err.response['Error']['Message'])
raise
else:
return response
```
Anda dapat memodifikasi klausa filter untuk hanya menampilkan hasil antara waktu tertentu atau berjalan terhadap tabel tertentu.
**Untuk menghentikan proses AWS Glue Data Quality yang sedang berlangsung:**
```
class GlueWrapper:
"""Encapsulates AWS Glue actions."""
def __init__(self, glue_client):
"""
:param glue_client: A Boto3 AWS Glue client.
"""
self.glue_client = glue_client
def cancel_data_quality_ruleset_evaluation_run(self, result_id):
"""
Cancels a given AWS Glue Data Quality run
:param result_id: The result id of a AWS Glue Data Quality run to cancel
"""
try:
response = self.client.cancel_data_quality_ruleset_evaluation_run(
ResultId=result_id
)
except ClientError as err:
logger.error(
"Couldn't cancel the AWS Glue Data Quality run. Here's why: %s: %s",
err.response['Error']['Code'], err.response['Error']['Message'])
raise
else:
return response
```
## Bekerja dengan hasil Kualitas Data AWS Glue
**Untuk mendapatkan hasil AWS Glue Data Quality run Anda:**
```
class GlueWrapper:
"""Encapsulates AWS Glue actions."""
def __init__(self, glue_client):
"""
:param glue_client: A Boto3 AWS Glue client.
"""
self.glue_client = glue_client
def get_data_quality_result(self, result_id):
"""
Outputs the result of an AWS Glue Data Quality Result
:param result_id: The result id of an AWS Glue Data Quality run
"""
try:
response = self.client.get_data_quality_result(
ResultId=result_id
)
except ClientError as err:
logger.error(
"Couldn't get the AWS Glue Data Quality result. Here's why: %s: %s",
err.response['Error']['Code'], err.response['Error']['Message'])
raise
else:
return response
```
**Untuk melihat statistik yang dikumpulkan untuk hasil kualitas data tertentu:**
```
import boto3
from botocore.exceptions import ClientError
import logging
logger = logging.getLogger(__name__)
class GlueWrapper:
"""Encapsulates AWS Glue actions."""
def __init__(self, glue_client):
"""
:param glue_client: A Boto3 AWS Glue client.
"""
self.glue_client = glue_client
def get_profile_for_data_quality_result(self, result_id):
"""
Outputs the statistic profile for a AWS Glue Data Quality Result
:param result_id: The result id of a AWS Glue Data Quality run
"""
try:
response = self.glue_client.get_data_quality_result(
ResultId=result_id
)
# the profile contains all statistics gathered for the result
profile_id = response['ProfileId']
profile = self.glue_client.list_data_quality_statistics(
ProfileId = profile_id
)
return profile
except ClientError as err:
logger.error(
"Couldn't retrieve Data Quality profile. Here's why: %s: %s",
err.response['Error']['Code'], err.response['Error']['Message'])
raise
```
**Untuk melihat rentang waktu untuk statistik yang dikumpulkan di beberapa proses kualitas data:**
```
class GlueWrapper:
"""Encapsulates AWS Glue actions."""
def __init__(self, glue_client):
"""
:param glue_client: A Boto3 AWS Glue client.
"""
self.glue_client = glue_client
def get_statistics_for_data_quality_result(self, profile_id):
"""
Outputs an array of datapoints for each statistic in the input result.
:param result_id: The profile id of a AWS Glue Data Quality run
"""
try:
profile = self.glue_client.list_data_quality_statistics(
ProfileId = profile_id
)
statistics = [self.glue_client.list_data_quality_statistics(
StatisticId = s['StatisticId']
) for s in profile['Statistics']]
return statistics
except ClientError as err:
logger.error(
"Couldn't retrieve Data Quality statistics. Here's why: %s: %s",
err.response['Error']['Code'], err.response['Error']['Message'])
raise
```
**Untuk melihat model deteksi anomali untuk statistik tertentu:**
```
class GlueWrapper:
"""Encapsulates AWS Glue actions."""
def __init__(self, glue_client):
"""
:param glue_client: A Boto3 AWS Glue client.
"""
self.glue_client = glue_client
def get_model_training_result_for_statistic(self, statistic_id, profile_id):
"""
Outputs the details (bounds) of anomaly detection training for the given statistic at the given profile.
:param statistic_id the model's statistic (the timeseries it is tracking)
:param profile_id the profile associated with the model (a point in the timeseries)
"""
try:
model = self.glue_client.get_data_quality_model_result(
ProfileId = profile_id, StatisticId = statistic_id
)
return model
except ClientError as err:
logger.error(
"Couldn't retrieve Data Quality model results. Here's why: %s: %s",
err.response['Error']['Code'], err.response['Error']['Message'])
raise
```
**Untuk mengecualikan titik data dari garis dasar deteksi anomali model statistiknya:**
```
class GlueWrapper:
"""Encapsulates AWS Glue actions."""
def __init__(self, glue_client):
"""
:param glue_client: A Boto3 AWS Glue client.
"""
self.glue_client = glue_client
def apply_exclusions_to_statistic(self, statistic_id, profile_ids):
"""
Annotate some points along a given statistic timeseries.
This example excludes the provided values; INCLUDE can also be used to undo this action.
:param statistic_id the statistic timeseries to annotate
:param profile_id the profiles we want to exclude (points in the timeseries)
"""
try:
response = self.glue_client.batch_put_data_quality_statistic_annotation(
InclusionAnnotations = [
{'ProfileId': prof_id,
'StatisticId': statistic_id,
'InclusionAnnotation': 'EXCLUDE'} for prof_id in profile_ids
]
)
return response['FailedInclusionAnnotations']
except ClientError as err:
logger.error(
"Couldn't store Data Quality annotations. Here's why: %s: %s",
err.response['Error']['Code'], err.response['Error']['Message'])
raise
```
**Untuk melihat status pelatihan model deteksi anomali untuk statistik tertentu:**
```
class GlueWrapper:
"""Encapsulates AWS Glue actions."""
def __init__(self, glue_client):
"""
:param glue_client: A Boto3 AWS Glue client.
"""
self.glue_client = glue_client
def get_model_training_status_for_statistic(self, statistic_id, profile_id):
"""
Outputs the status of anomaly detection training for the given statistic at the given profile.
:param statistic_id the model's statistic (the timeseries it is tracking)
:param profile_id the profile associated with the model (a point in the timeseries)
"""
try:
model = self.glue_client.get_data_quality_model(
ProfileId = profile_id, StatisticId = statistic_id
)
return model
except ClientError as err:
logger.error(
"Couldn't retrieve Data Quality statistics. Here's why: %s: %s",
err.response['Error']['Code'], err.response['Error']['Message'])
raise
```
**Untuk mengecualikan semua hasil dari kualitas data tertentu, jalankan dari garis dasar deteksi anomali:**
```
class GlueWrapper:
"""Encapsulates AWS Glue actions."""
def __init__(self, glue_client):
"""
:param glue_client: A Boto3 AWS Glue client.
"""
self.glue_client = glue_client
def apply_exclusions_to_profile(self, profile_id):
"""
Exclude datapoints produced by a run across statistic timeseries.
This example excludes the provided values; INCLUDE can also be used to undo this action.
:param profile_id the profiles we want to exclude (points in the timeseries)
"""
try:
response = self.glue_client.put_data_quality_profile_annotation(
ProfileId = profile_id,
InclusionAnnotation = "EXCLUDE"
)
return response
except ClientError as err:
logger.error(
"Couldn't store Data Quality annotations. Here's why: %s: %s",
err.response['Error']['Code'], err.response['Error']['Message'])
raise
```
**Untuk mendapatkan hasil dari kualitas data yang diberikan, jalankan dan tampilkan hasilnya:**
Dengan Kualitas Data AWS Glue`runID`, Anda dapat mengekstrak `resultID` untuk kemudian mendapatkan hasil aktual, seperti yang ditunjukkan di bawah ini:
```
response = client.get_data_quality_ruleset_evaluation_run(
RunId='dqrun-abca77ee126abe1378c1da1ae0750d7dxxxx'
)
resultID = response['ResultIds'][0]
response = client.get_data_quality_result(
ResultId=resultID
)
print(resp['RuleResults'])
```