Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add/data quality #113

Merged
merged 3 commits into from
Nov 28, 2023
Merged

Add/data quality #113

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
7575557
Add data/quality/{id} endpoint
PGijsbers Nov 27, 2023
5e737b7
Add authentication check, document unsupported error messaging
PGijsbers Nov 28, 2023
dcb93cd
Test failure cases and document them in separate test
PGijsbers Nov 28, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
54 changes: 49 additions & 5 deletions src/routers/v1/qualities.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,10 +1,14 @@
from typing import Annotated, Literal
import http.client
from typing import Annotated, Any, Literal

from database.datasets import list_all_qualities
from fastapi import APIRouter, Depends
from sqlalchemy import Connection
from database.datasets import get_dataset, list_all_qualities
from database.users import User, UserGroup
from fastapi import APIRouter, Depends, HTTPException
from schemas.datasets.openml import Quality
from sqlalchemy import Connection, text

from routers.dependencies import expdb_connection
from routers.dependencies import expdb_connection, fetch_user
from routers.v2.datasets import DatasetError

router = APIRouter(prefix="/v1/datasets", tags=["datasets"])

Expand All @@ -19,3 +23,43 @@ def list_qualities(
"quality": qualities,
},
}


def _user_can_see_dataset(dataset: dict[str, Any], user: User) -> bool:
if dataset["visibility"] == "public":
return True
return user is not None and (
dataset["uploader"] == user.user_id or UserGroup.ADMIN in user.groups
)


@router.get("/qualities/{dataset_id}")
def get_qualities(
dataset_id: int,
user: Annotated[User, Depends(fetch_user)],
expdb: Annotated[Connection, Depends(expdb_connection)],
) -> list[Quality]:
dataset = get_dataset(dataset_id, expdb)
if not dataset or not _user_can_see_dataset(dataset, user):
raise HTTPException(
status_code=http.client.PRECONDITION_FAILED,
detail={"code": DatasetError.NO_DATA_FILE, "message": "Unknown dataset"},
) from None
rows = expdb.execute(
text(
"""
SELECT `quality`,`value`
FROM data_quality
WHERE `data`=:dataset_id
""",
),
parameters={"dataset_id": dataset_id},
)
return [Quality(name=row.quality, value=row.value) for row in rows]
# The PHP API provided (sometime) helpful error messages
# if not qualities:
# check if dataset exists: error 360
# check if user has access: error 361
# check if there is a data processed entry and forward the error: 364
# if nothing in process table: 363
# otherwise: error 362
5 changes: 5 additions & 0 deletions src/schemas/datasets/openml.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,6 +24,11 @@ class DatasetStatus(StrEnum):
IN_PREPARATION = "in_preparation"


class Quality(BaseModel):
name: str
value: float | None


class DatasetMetadata(BaseModel):
id_: int = Field(json_schema_extra={"example": 1}, alias="id")
visibility: Visibility = Field(json_schema_extra={"example": Visibility.PUBLIC})
Expand Down
153 changes: 153 additions & 0 deletions tests/routers/v1/qualities_test.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -160,3 +160,156 @@ def test_list_qualities(api_client: TestClient, expdb_test: Connection) -> None:
response = api_client.get("/v1/datasets/qualities/list")
assert response.status_code == http.client.OK
assert expected == response.json()


def test_get_quality(api_client: TestClient) -> None:
response = api_client.get("/v1/datasets/qualities/1")
assert response.status_code == http.client.OK
expected = [
{"name": "AutoCorrelation", "value": 0.6064659977703456},
{"name": "CfsSubsetEval_DecisionStumpAUC", "value": 0.9067742570970945},
{"name": "CfsSubsetEval_DecisionStumpErrRate", "value": 0.13251670378619154},
{"name": "CfsSubsetEval_DecisionStumpKappa", "value": 0.6191022730108037},
{"name": "CfsSubsetEval_NaiveBayesAUC", "value": 0.9067742570970945},
{"name": "CfsSubsetEval_NaiveBayesErrRate", "value": 0.13251670378619154},
{"name": "CfsSubsetEval_NaiveBayesKappa", "value": 0.6191022730108037},
{"name": "CfsSubsetEval_kNN1NAUC", "value": 0.9067742570970945},
{"name": "CfsSubsetEval_kNN1NErrRate", "value": 0.13251670378619154},
{"name": "CfsSubsetEval_kNN1NKappa", "value": 0.6191022730108037},
{"name": "ClassEntropy", "value": 1.189833856204398},
{"name": "DecisionStumpAUC", "value": 0.8652735384332186},
{"name": "DecisionStumpErrRate", "value": 0.22828507795100222},
{"name": "DecisionStumpKappa", "value": 0.4503332218612649},
{"name": "Dimensionality", "value": 0.043429844097995544},
{"name": "EquivalentNumberOfAtts", "value": 26.839183802676523},
{"name": "J48.00001.AUC", "value": 0.9391585368767195},
{"name": "J48.00001.ErrRate", "value": 0.10356347438752785},
{"name": "J48.00001.Kappa", "value": 0.7043302166347443},
{"name": "J48.0001.AUC", "value": 0.9391585368767195},
{"name": "J48.0001.ErrRate", "value": 0.10356347438752785},
{"name": "J48.0001.Kappa", "value": 0.7043302166347443},
{"name": "J48.001.AUC", "value": 0.9391585368767195},
{"name": "J48.001.ErrRate", "value": 0.10356347438752785},
{"name": "J48.001.Kappa", "value": 0.7043302166347443},
{"name": "MajorityClassPercentage", "value": 76.16926503340757},
{"name": "MajorityClassSize", "value": 684.0},
{"name": "MaxAttributeEntropy", "value": 1.8215224482924186},
{"name": "MaxKurtosisOfNumericAtts", "value": 13.215477213878724},
{"name": "MaxMeansOfNumericAtts", "value": 1263.0946547884187},
{"name": "MaxMutualInformation", "value": 0.40908953764451},
{"name": "MaxNominalAttDistinctValues", "value": 7.0},
{"name": "MaxSkewnessOfNumericAtts", "value": 3.7616019689156888},
{"name": "MaxStdDevOfNumericAtts", "value": 1871.3991072665933},
{"name": "MeanAttributeEntropy", "value": 0.2515351603742048},
{"name": "MeanKurtosisOfNumericAtts", "value": 4.6480244352098286},
{"name": "MeanMeansOfNumericAtts", "value": 348.50426818856715},
{"name": "MeanMutualInformation", "value": 0.044331968697414056},
{"name": "MeanNoiseToSignalRatio", "value": 4.673900071775454},
{"name": "MeanNominalAttDistinctValues", "value": 1.6363636363636362},
{"name": "MeanSkewnessOfNumericAtts", "value": 2.0269825910719437},
{"name": "MeanStdDevOfNumericAtts", "value": 405.17326983791025},
{"name": "MinAttributeEntropy", "value": -0.0},
{"name": "MinKurtosisOfNumericAtts", "value": -0.9723842038435437},
{"name": "MinMeansOfNumericAtts", "value": 1.1985489977728285},
{"name": "MinMutualInformation", "value": 0.0},
{"name": "MinNominalAttDistinctValues", "value": 0.0},
{"name": "MinSkewnessOfNumericAtts", "value": 0.07299048442083138},
{"name": "MinStdDevOfNumericAtts", "value": 0.871208280971892},
{"name": "MinorityClassPercentage", "value": 0.8908685968819599},
{"name": "MinorityClassSize", "value": 8.0},
{"name": "NaiveBayesAUC", "value": 0.9315907109421729},
{"name": "NaiveBayesErrRate", "value": 0.24610244988864144},
{"name": "NaiveBayesKappa", "value": 0.5569590016631507},
{"name": "NumberOfBinaryFeatures", "value": 4.0},
{"name": "NumberOfClasses", "value": 5.0},
{"name": "NumberOfFeatures", "value": 39.0},
{"name": "NumberOfInstances", "value": 898.0},
{"name": "NumberOfInstancesWithMissingValues", "value": 898.0},
{"name": "NumberOfMissingValues", "value": 22175.0},
{"name": "NumberOfNumericFeatures", "value": 6.0},
{"name": "NumberOfSymbolicFeatures", "value": 33.0},
{"name": "PercentageOfBinaryFeatures", "value": 10.256410256410255},
{"name": "PercentageOfInstancesWithMissingValues", "value": 100.0},
{"name": "PercentageOfMissingValues", "value": 63.317343384158534},
{"name": "PercentageOfNumericFeatures", "value": 15.384615384615385},
{"name": "PercentageOfSymbolicFeatures", "value": 84.61538461538461},
{"name": "Quartile1AttributeEntropy", "value": 0.0},
{"name": "Quartile1KurtosisOfNumericAtts", "value": -0.40305022089010156},
{"name": "Quartile1MeansOfNumericAtts", "value": 3.025695155902005},
{"name": "Quartile1MutualInformation", "value": 0.0},
{"name": "Quartile1SkewnessOfNumericAtts", "value": 0.967384603629726},
{"name": "Quartile1StdDevOfNumericAtts", "value": 10.505435772171138},
{"name": "Quartile2AttributeEntropy", "value": 0.0},
{"name": "Quartile2KurtosisOfNumericAtts", "value": 1.6372437439142264},
{"name": "Quartile2MeansOfNumericAtts", "value": 21.222160356347437},
{"name": "Quartile2MutualInformation", "value": 0.0},
{"name": "Quartile2SkewnessOfNumericAtts", "value": 1.6547313364025702},
{"name": "Quartile2StdDevOfNumericAtts", "value": 69.85338529046133},
{"name": "Quartile3AttributeEntropy", "value": 0.2385631077559124},
{"name": "Quartile3KurtosisOfNumericAtts", "value": 12.741748058445403},
{"name": "Quartile3MeansOfNumericAtts", "value": 901.2636692650334},
{"name": "Quartile3MutualInformation", "value": 0.0206465881071925},
{"name": "Quartile3SkewnessOfNumericAtts", "value": 3.7546438249219056},
{"name": "Quartile3StdDevOfNumericAtts", "value": 771.8590427889504},
{"name": "REPTreeDepth1AUC", "value": 0.962680369298288},
{"name": "REPTreeDepth1ErrRate", "value": 0.08463251670378619},
{"name": "REPTreeDepth1Kappa", "value": 0.768583383630482},
{"name": "REPTreeDepth2AUC", "value": 0.962680369298288},
{"name": "REPTreeDepth2ErrRate", "value": 0.08463251670378619},
{"name": "REPTreeDepth2Kappa", "value": 0.768583383630482},
{"name": "REPTreeDepth3AUC", "value": 0.962680369298288},
{"name": "REPTreeDepth3ErrRate", "value": 0.08463251670378619},
{"name": "REPTreeDepth3Kappa", "value": 0.768583383630482},
{"name": "RandomTreeDepth1AUC", "value": 0.9296999989655875},
{"name": "RandomTreeDepth1ErrRate", "value": 0.0801781737193764},
{"name": "RandomTreeDepth1Kappa", "value": 0.7953250436852635},
{"name": "RandomTreeDepth2AUC", "value": 0.9296999989655875},
{"name": "RandomTreeDepth2ErrRate", "value": 0.0801781737193764},
{"name": "RandomTreeDepth2Kappa", "value": 0.7953250436852635},
{"name": "RandomTreeDepth3AUC", "value": 0.9296999989655875},
{"name": "RandomTreeDepth3ErrRate", "value": 0.0801781737193764},
{"name": "RandomTreeDepth3Kappa", "value": 0.7953250436852635},
{"name": "StdvNominalAttDistinctValues", "value": 1.5576059718800395},
{"name": "kNN1NAUC", "value": 0.8721948540771287},
{"name": "kNN1NErrRate", "value": 0.06347438752783964},
{"name": "kNN1NKappa", "value": 0.8261102938928316},
]
assert response.json() == expected


@pytest.mark.php()
@pytest.mark.parametrize(
"data_id",
list(set(range(1, 132)) - {55, 56, 59, 116, 130}),
)
def test_get_quality_identical(data_id: int, api_client: TestClient) -> None:
php_response = httpx.get(f"http://server-api-php-api-1:80/api/v1/json/data/qualities/{data_id}")
python_response = api_client.get(f"/v1/datasets/qualities/{data_id}")
assert python_response.status_code == php_response.status_code

expected = [
{
"name": quality["name"],
"value": None if quality["value"] == [] else float(quality["value"]),
}
for quality in php_response.json()["data_qualities"]["quality"]
]
assert python_response.json() == expected


@pytest.mark.php()
@pytest.mark.parametrize(
"data_id",
[55, 56, 59, 116, 130, 132],
)
def test_get_quality_identical_error(data_id: int, api_client: TestClient) -> None:
if data_id in [55, 56, 59]:
pytest.skip("Detailed error for code 364 (failed processing) not yet supported.")
if data_id in [116]:
pytest.skip("Detailed error for code 362 (no qualities) not yet supported.")
php_response = httpx.get(f"http://server-api-php-api-1:80/api/v1/json/data/qualities/{data_id}")
python_response = api_client.get(f"/v1/datasets/qualities/{data_id}")
assert python_response.status_code == php_response.status_code
# The "dataset unknown" error currently has a separate code in PHP depending on
# where it occurs (e.g., get dataset->113 get quality->361)
assert python_response.json()["detail"]["message"] == php_response.json()["error"]["message"]