Merge branch 'main' of github.com:radicalbit/radicalbit-ai-monitoring…

… into feature/ROS-310-add-latest-current-and-reference-job-status-to-modelout

# Conflicts:
#	api/app/models/model_dto.py

.github/workflows/authors.yaml

@@ -0,0 +1,21 @@
+name: Update Authors
+on:
+  push:
+    branches:
+      - main
+jobs:
+  update-authors:
+    runs-on: ubuntu-22.04
+    steps:
+    - uses: actions/checkout@v4
+    - name: Run update-authors.sh script file
+      run: |
+         chmod +x ./tools/update-authors.sh
+         ./tools/update-authors.sh
+      shell: bash
+    - name: Commit changes
+      uses: stefanzweifel/git-auto-commit-action@v5
+      with:
+        commit_message: Update AUTHORS.md file
+      env:
+        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

AUTHORS.md

@@ -0,0 +1,35 @@
+<!--
+Copyright (c) radicalbit-ai-monitoring contributors.
+Licensed under https://github.com/radicalbit/radicalbit-ai-monitoring/blob/main/LICENSE
+-->
+
+Contributors to radicalbit-ai-monitoring
+============================
+radicalbit-ai-monitoring is developed and maintained by a community of people interested in providing a comprehensive solution for monitoring your Machine Learning and Large Language Models in production.
+
+
+<p align="center">
+<img src="https://contributors-img.web.app/image?repo=radicalbit/radicalbit-ai-monitoring" width="500"/>
+</p>
+
+
+Contributors (ordered by first contribution.)
+-------------------------------------
+[Full List of Contributors](https://github.com/radicalbit/radicalbit-ai-monitoring/graphs/contributors)
+
+- Paolo Filippelli (<[email protected]>)
+- Roberto Bentivoglio (<[email protected]>)
+- Luca Tagliabue (<[email protected]>)
+- Mauro Cortellazzi (<[email protected]>)
+- Stefano Zamboni (<[email protected]>)
+- mmariniello90 (<[email protected]>)
+- Davide Valleri (<[email protected]>)
+- Lorenzo D'Agostino (<[email protected]>)
+- Daniele Tria (<[email protected]>)
+- rivamarco (<[email protected]>)
+- d-croci (<[email protected]>)
+- cybermarinella (<[email protected]>)
+- rivamarco (<[email protected]>)
+- BigMoby (<[email protected]>)
+
+ #### Generated by tools/update-authors.sh.

api/README.md

@@ -33,6 +33,18 @@ to check.
 
 ## Test
 
+Please install a PostgreSQL database locally. For example, on a macOS platform, execute:
+
+```bash
+brew install postgresql
+```
+
+Note: If any errors occur during pytest runs, please stop the local database service by executing:
+
+```bash
+brew services stop postgresql
+```
+
 Tests are done with `pytest`
 
 Run

api/app/models/model_dto.py

@@ -1,12 +1,14 @@
 import datetime
 from enum import Enum
-from typing import List, Optional
+from typing import List, Optional, Self
 from uuid import UUID
 
-from pydantic import BaseModel, ConfigDict
+from pydantic import BaseModel, ConfigDict, model_validator
 from pydantic.alias_generators import to_camel
 
 from app.db.dao.model_dao import Model
+from app.models.inferred_schema_dto import SupportedTypes
+from app.models.utils import is_none, is_number, is_number_or_string, is_optional_float
 from app.db.dao.reference_dataset_dao import ReferenceDataset
 from app.db.dao.current_dataset_dao import CurrentDataset
 from app.models.job_status import JobStatus
@@ -31,15 +33,15 @@ class Granularity(str, Enum):
     MONTH = 'MONTH'
 
 
-class ColumnDefinition(BaseModel):
+class ColumnDefinition(BaseModel, validate_assignment=True):
     name: str
-    type: str
+    type: SupportedTypes
 
     def to_dict(self):
         return self.model_dump()
 
 
-class OutputType(BaseModel):
+class OutputType(BaseModel, validate_assignment=True):
     prediction: ColumnDefinition
     prediction_proba: Optional[ColumnDefinition] = None
     output: List[ColumnDefinition]
@@ -50,7 +52,7 @@ def to_dict(self):
         return self.model_dump()
 
 
-class ModelIn(BaseModel):
+class ModelIn(BaseModel, validate_assignment=True):
     name: str
     description: Optional[str] = None
     model_type: ModelType
@@ -67,6 +69,74 @@ class ModelIn(BaseModel):
         populate_by_name=True, alias_generator=to_camel, protected_namespaces=()
     )
 
+    @model_validator(mode='after')
+    def validate_target(self) -> Self:
+        checked_model_type: ModelType = self.model_type
+        match checked_model_type:
+            case ModelType.BINARY:
+                if not is_number(self.target.type):
+                    raise ValueError(
+                        f'target must be a number for a ModelType.BINARY, has been provided [{self.target}]'
+                    )
+                return self
+            case ModelType.MULTI_CLASS:
+                if not is_number_or_string(self.target.type):
+                    raise ValueError(
+                        f'target must be a number or string for a ModelType.MULTI_CLASS, has been provided [{self.target}]'
+                    )
+                return self
+            case ModelType.REGRESSION:
+                if not is_number(self.target.type):
+                    raise ValueError(
+                        f'target must be a number for a ModelType.REGRESSION, has been provided [{self.target}]'
+                    )
+                return self
+            case _:
+                raise ValueError('not supported type for model_type')
+
+    @model_validator(mode='after')
+    def validate_outputs(self) -> Self:
+        checked_model_type: ModelType = self.model_type
+        match checked_model_type:
+            case ModelType.BINARY:
+                if not is_number(self.outputs.prediction.type):
+                    raise ValueError(
+                        f'prediction must be a number for a ModelType.BINARY, has been provided [{self.outputs.prediction}]'
+                    )
+                if not is_optional_float(self.outputs.prediction_proba.type):
+                    raise ValueError(
+                        f'prediction_proba must be an optional float for a ModelType.BINARY, has been provided [{self.outputs.prediction_proba}]'
+                    )
+                return self
+            case ModelType.MULTI_CLASS:
+                if not is_number_or_string(self.outputs.prediction.type):
+                    raise ValueError(
+                        f'prediction must be a number or string for a ModelType.MULTI_CLASS, has been provided [{self.outputs.prediction}]'
+                    )
+                if not is_optional_float(self.outputs.prediction_proba.type):
+                    raise ValueError(
+                        f'prediction_proba must be an optional float for a ModelType.MULTI_CLASS, has been provided [{self.outputs.prediction_proba}]'
+                    )
+                return self
+            case ModelType.REGRESSION:
+                if not is_number(self.outputs.prediction.type):
+                    raise ValueError(
+                        f'prediction must be a number for a ModelType.REGRESSION, has been provided [{self.outputs.prediction}]'
+                    )
+                if not is_none(self.outputs.prediction_proba.type):
+                    raise ValueError(
+                        f'prediction_proba must be None for a ModelType.REGRESSION, has been provided [{self.outputs.prediction_proba}]'
+                    )
+                return self
+            case _:
+                raise ValueError('not supported type for model_type')
+
+    @model_validator(mode='after')
+    def timestamp_must_be_datetime(self) -> Self:
+        if not self.timestamp.type == SupportedTypes.datetime:
+            raise ValueError('timestamp must be a datetime')
+        return self
+
     def to_model(self) -> Model:
         now = datetime.datetime.now(tz=datetime.UTC)
         return Model(

api/app/models/utils.py

@@ -0,0 +1,19 @@
+from typing import Any, Optional
+
+from app.models.inferred_schema_dto import SupportedTypes
+
+
+def is_number(value: SupportedTypes):
+    return value in (SupportedTypes.int, SupportedTypes.float)
+
+
+def is_number_or_string(value: SupportedTypes):
+    return value in (SupportedTypes.int, SupportedTypes.float, SupportedTypes.string)
+
+
+def is_optional_float(value: Optional[SupportedTypes] = None) -> bool:
+    return value in (None, SupportedTypes.float)
+
+
+def is_none(value: Any) -> bool:
+    return value is None

api/tests/commons/db_mock.py

@@ -8,7 +8,15 @@
 from app.db.tables.reference_dataset_metrics_table import ReferenceDatasetMetrics
 from app.db.tables.reference_dataset_table import ReferenceDataset
 from app.models.job_status import JobStatus
-from app.models.model_dto import DataType, Granularity, ModelIn, ModelType
+from app.models.model_dto import (
+    ColumnDefinition,
+    DataType,
+    Granularity,
+    ModelIn,
+    ModelType,
+    OutputType,
+    SupportedTypes,
+)
 
 MODEL_UUID = uuid.uuid4()
 REFERENCE_UUID = uuid.uuid4()
@@ -26,7 +34,7 @@ def get_sample_model(
     features: List[Dict] = [{'name': 'feature1', 'type': 'string'}],
     outputs: Dict = {
         'prediction': {'name': 'pred1', 'type': 'int'},
-        'prediction_proba': {'name': 'prob1', 'type': 'double'},
+        'prediction_proba': {'name': 'prob1', 'type': 'float'},
         'output': [{'name': 'output1', 'type': 'string'}],
     },
     target: Dict = {'name': 'target1', 'type': 'string'},
@@ -59,14 +67,20 @@ def get_sample_model_in(
     model_type: str = ModelType.BINARY.value,
     data_type: str = DataType.TEXT.value,
     granularity: str = Granularity.DAY.value,
-    features: List[Dict] = [{'name': 'feature1', 'type': 'string'}],
-    outputs: Dict = {
-        'prediction': {'name': 'pred1', 'type': 'int'},
-        'prediction_proba': {'name': 'prob1', 'type': 'double'},
-        'output': [{'name': 'output1', 'type': 'string'}],
-    },
-    target: Dict = {'name': 'target1', 'type': 'string'},
-    timestamp: Dict = {'name': 'timestamp', 'type': 'datetime'},
+    features: List[ColumnDefinition] = [
+        ColumnDefinition(name='feature1', type=SupportedTypes.string)
+    ],
+    outputs: OutputType = OutputType(
+        prediction=ColumnDefinition(name='pred1', type=SupportedTypes.int),
+        prediction_proba=ColumnDefinition(name='prob1', type=SupportedTypes.float),
+        output=[ColumnDefinition(name='output1', type=SupportedTypes.string)],
+    ),
+    target: ColumnDefinition = ColumnDefinition(
+        name='target1', type=SupportedTypes.int
+    ),
+    timestamp: ColumnDefinition = ColumnDefinition(
+        name='timestamp', type=SupportedTypes.datetime
+    ),
     frameworks: Optional[str] = None,
     algorithm: Optional[str] = None,
 ):

api/tests/commons/modelin_factory.py

@@ -0,0 +1,68 @@
+from app.models.inferred_schema_dto import SupportedTypes
+from app.models.model_dto import (
+    ColumnDefinition,
+    DataType,
+    Granularity,
+    ModelType,
+    OutputType,
+)
+
+
+def get_model_sample_wrong(fail_field: str, model_type: ModelType):
+    prediction = None
+    prediction_proba = None
+    if fail_field == 'outputs.prediction' and model_type == ModelType.BINARY:
+        prediction = ColumnDefinition(name='pred1', type=SupportedTypes.string)
+    elif fail_field == 'outputs.prediction' and model_type == ModelType.MULTI_CLASS:
+        prediction = ColumnDefinition(name='pred1', type=SupportedTypes.datetime)
+    elif fail_field == 'outputs.prediction' and model_type == ModelType.REGRESSION:
+        prediction = ColumnDefinition(name='pred1', type=SupportedTypes.string)
+    else:
+        prediction = ColumnDefinition(name='pred1', type=SupportedTypes.int)
+
+    if (
+        fail_field == 'outputs.prediction_proba'
+        and model_type == ModelType.BINARY
+        or fail_field == 'outputs.prediction_proba'
+        and model_type == ModelType.MULTI_CLASS
+    ):
+        prediction_proba = ColumnDefinition(name='prob1', type=SupportedTypes.int)
+    elif (
+        fail_field == 'outputs.prediction_proba' and model_type == ModelType.REGRESSION
+    ):
+        prediction_proba = ColumnDefinition(name='prob1', type=SupportedTypes.float)
+    else:
+        prediction_proba = ColumnDefinition(name='prob1', type=SupportedTypes.float)
+
+    target: ColumnDefinition = None
+    if fail_field == 'target' and model_type == ModelType.BINARY:
+        target = ColumnDefinition(name='target1', type=SupportedTypes.string)
+    elif fail_field == 'target' and model_type == ModelType.MULTI_CLASS:
+        target = ColumnDefinition(name='target1', type=SupportedTypes.datetime)
+    elif fail_field == 'target' and model_type == ModelType.REGRESSION:
+        target = ColumnDefinition(name='target1', type=SupportedTypes.string)
+    else:
+        target = ColumnDefinition(name='target1', type=SupportedTypes.int)
+
+    timestamp: ColumnDefinition = None
+    if fail_field == 'timestamp':
+        timestamp = ColumnDefinition(name='timestamp', type=SupportedTypes.string)
+    else:
+        timestamp = ColumnDefinition(name='timestamp', type=SupportedTypes.datetime)
+
+    return {
+        'name': 'model_name',
+        'model_type': model_type,
+        'data_type': DataType.TEXT,
+        'granularity': Granularity.DAY,
+        'features': [ColumnDefinition(name='feature1', type=SupportedTypes.string)],
+        'outputs': OutputType(
+            prediction=prediction,
+            prediction_proba=prediction_proba,
+            output=[ColumnDefinition(name='output1', type=SupportedTypes.string)],
+        ),
+        'target': target,
+        'timestamp': timestamp,
+        'frameworks': None,
+        'algorithm': None,
+    }

api/tests/validation/__init__.py

@@ -0,0 +1 @@
+