feat: model training & selection

.editorconfig

@@ -30,3 +30,6 @@ indent_size = unset
 
 [*.txt]
 indent_size = unset
+
+[*.ipynb]
+indent_size = unset

.env

@@ -3,3 +3,4 @@ POSTGRES_USER='guest'
 POSTGRES_PASSWORD=')M8z*yss$cRxw7(&'
 
 BACKEND_DATABASE__URI = 'postgresql://guest:)M8z*yss$cRxw7(&@database:5432/heartbeat'
+BACKEND_CHECKPOINT_PATH = './model.joblib'

.gitignore

@@ -152,3 +152,7 @@ Thumbs.db #thumbnail cache on Windows
 # End of https://www.toptal.com/developers/gitignore/api/python
 
 *.csv
+*.pkl
+*.joblib
+
+.vscode

.pre-commit-config.yaml

@@ -88,6 +88,12 @@ repos:
       - id: poetry-check
       - id: poetry-lock
         args: [--check]
+  - repo: https://github.com/nbQA-dev/nbQA
+    rev: 1.8.5
+    hooks:
+      - id: nbqa-check-ast
+      - id: nbqa-ruff
+      - id: nbqa-mypy
   - repo: meta
     hooks:
       - id: check-hooks-apply

README.md

@@ -45,6 +45,14 @@
   </a>
 </p>
 
+Cardiovascular diseases (CVDs) are the number 1 cause of death globally, taking an estimated 17.9 million lives each year, which accounts for 31% of all deaths worldwide. Four out of 5CVD deaths are due to heart attacks and strokes, and one-third of these deaths occur prematurely in people under 70 years of age. Heart failure is a common event caused by CVDs and this dataset contains 11 features that can be used to predict a possible heart disease.
+
+People with cardiovascular disease or who are at high cardiovascular risk (due to the presence of one or more risk factors such as hypertension, diabetes, hyperlipidaemia or already established disease) need early detection and management wherein a machine learning model can be of great help.
+
+> Source: [https://www.kaggle.com/datasets/fedesoriano/heart-failure-prediction](https://www.kaggle.com/datasets/fedesoriano/heart-failure-prediction)
+
+In this project, we create a complete solution featuring a [`FastAPI`](https://fastapi.tiangolo.com/) backend and a [`React`](https://react.dev/) frontend. We perform Exploratory Data Analysis (EDA) and develop a machine learning model using [`scikit-learn`](https://scikit-learn.org).
+
 ## :rocket: Running the project
 
 > [!IMPORTANT]
@@ -65,6 +73,7 @@
     POSTGRES_PASSWORD=')M8z*yss$cRxw7(&'
 
     BACKEND_DATABASE__URI = 'postgresql://guest:)M8z*yss$cRxw7(&@database:5432/heartbeat'
+    BACKEND_CHECKPOINT_PATH = './model.joblib'
     ```
 
 3. **Build and start the services** using Docker Compose.
@@ -104,6 +113,7 @@ Running the backend:
 
 ```shell
 export BACKEND_DATABASE__URI='postgresql://guest:)M8z*yss$cRxw7(&@localhost:5432/heartbeat'
+export BACKEND_CHECKPOINT_PATH='./model.joblib'
 poetry env use 3.11
 poetry install
 python src/api/app.py
@@ -126,8 +136,6 @@ npm run dev -- --host --port 80
 
 ## :computer: Deploying to production
 
-## Deploying to Production
-
 Deploying to production involves several crucial steps, including setting up a server, configuring DNS, and managing SSL certificates. [`Traefik`](https://github.com/traefik/traefik) simplifies many of these tasks, acting as a powerful reverse proxy and load balancer. For a comprehensive guide on deploying your FastAPI application with Traefik, read the full article [here](https://github.com/tiangolo/blog-posts/blob/master/deploying-fastapi-apps-with-https-powered-by-traefik/README.md).
 
 ## :bookmark_tabs: Contributing

docker-compose.yaml

@@ -33,6 +33,8 @@ services:
       interval: 5s
       timeout: 10s
       retries: 10
+    volumes:
+      - ${PWD}/model.joblib:/home/app/app/model.joblib
   database:
     env_file: .env
     container_name: database

pyproject.toml

@@ -44,6 +44,11 @@ pandas = "^2.2.2"
 seaborn = "^0.13.2"
 matplotlib = "^3.9.1"
 scikit-learn = "^1.5.1"
+xgboost = "^2.1.0"
+tqdm = "^4.66.4"
+jupyter = "^1.0.0"
+ipywidgets = "^8.1.3"
+joblib = "^1.4.2"
 
 [tool.poetry.group.dev.dependencies]
 mypy = "*"

src/api/app.py

@@ -1,7 +1,6 @@
 import uvicorn
 from fastapi import APIRouter, FastAPI
 from fastapi.middleware.cors import CORSMiddleware
-
 from api.controllers.heartbeat_controller import api as heartbeat_controller
 from api.controllers.monitor_controller import api as monitor_controller
 from api.settings import Settings
@@ -49,14 +48,9 @@ def register_middlewares(app: FastAPI) -> FastAPI:
     return app
 
 
-def register_events(app: FastAPI) -> FastAPI:
-    return app
-
-
 def create_app() -> FastAPI:
     app = initialize_api()
     app = register_configuration(app)
-    app = register_events(app)
     app = register_controllers(app)
     app = register_middlewares(app)
 

src/api/bot/__init__.py

@@ -1,6 +1,22 @@
 from api.schemas.heartbeat import HeartBeatCreateSchema
+from sklearn.pipeline import Pipeline
+import pandas as pd
 
 
-class Bot:
+class Bot(object):
+    def __init__(self, model: Pipeline) -> None:
+        self._model = model
+
     def predict(self, heartbeat: HeartBeatCreateSchema) -> bool:
-        return True
+        payload = {
+            "Age": heartbeat.age,
+            "Sex": heartbeat.sex.value,
+            "ChestPain": heartbeat.chest_pain_type.value,
+            "FastingBS": int(heartbeat.fasting_blood_sugar),
+            "MaxHR": heartbeat.max_heart_rate,
+            "ExerciseAngina": int(heartbeat.exercise_angina),
+            "Oldpeak": heartbeat.old_peak,
+            "ST_Slope": heartbeat.st_slope.value,
+        }
+
+        return self._model.predict(pd.DataFrame([payload]))[0]

src/api/dependencies.py

@@ -1,6 +1,6 @@
 from fastapi import Depends
 from fastapi.requests import Request
-
+import joblib
 from api.bot import Bot
 from api.repositories.heartbeat_repository import HeartBeatRepository
 from api.resources.database import Database
@@ -25,8 +25,10 @@ async def get_heartbeat_repository(
     return HeartBeatRepository(database)
 
 
-async def get_bot(request: Request):
-    return Bot()
+async def get_bot(request: Request, settings: Settings = Depends(get_settings)):
+    model = joblib.load(settings.checkpoint_path)
+
+    return Bot(model)
 
 
 async def get_heartbeat_service(

src/api/models/heartbeat.py

@@ -18,14 +18,6 @@ class ChestPain(IntEnum):
     ASYMPTOMATIC = auto()
 
 
-class RestingElectrocardiogram(IntEnum):
-    NORMAL = auto()
-    STT = auto()  # having ST-T wave abnormality (T wave inversions and/or ST elevation or depression of > 0.05 mV)
-    LVH = (
-        auto()
-    )  # showing probable or definite left ventricular hypertrophy by Estes' criteria
-
-
 class StSlope(IntEnum):
     UP = auto()
     FLAT = auto()
@@ -38,13 +30,7 @@ class HeartBeatModel(PkModel):
     age: Mapped[int] = Column(Integer, nullable=False)
     sex: Mapped[Sex] = Column(Enum(Sex), nullable=False)
     chest_pain_type: Mapped[ChestPain] = Column(Enum(ChestPain), nullable=False)
-    resting_blood_pressure: Mapped[int] = Column(Integer, nullable=False)
-    cholesterol: Mapped[int] = Column(Integer, nullable=False)
     fasting_blood_sugar: Mapped[bool] = Column(Boolean, nullable=False)
-    resting_electrocardiogram: Mapped[RestingElectrocardiogram] = Column(
-        Enum(RestingElectrocardiogram),
-        nullable=False,
-    )
     max_heart_rate: Mapped[int] = Column(Integer, nullable=False)
     exercise_angina: Mapped[bool] = Column(Boolean, nullable=False)
     old_peak: Mapped[float] = Column(Float, nullable=False)

src/api/repositories/heartbeat_repository.py

@@ -3,6 +3,8 @@
 
 from sqlalchemy import exc
 
+from sqlalchemy import desc
+
 from api.models.heartbeat import HeartBeatModel
 from api.repositories import Repository
 from api.resources.database import Database
@@ -22,7 +24,11 @@ def __init__(self, database: Database) -> None:
 
     def get_all(self) -> Iterator[HeartBeatModel]:
         with self.session_factory() as session:
-            return session.query(HeartBeatModel).all()
+            return (
+                session.query(HeartBeatModel)
+                .order_by(desc(HeartBeatModel.created_at))
+                .all()
+            )
 
     def get_by_id(self, id: str) -> HeartBeatModel:
         with self.session_factory() as session:

src/api/schemas/heartbeat.py

@@ -2,7 +2,6 @@
 
 from api.models.heartbeat import (
     ChestPain,
-    RestingElectrocardiogram,
     Sex,
     StSlope,
 )
@@ -15,18 +14,10 @@ class Config:
     age: int = Field(..., ge=0, le=130, description="Age of the patient [years]")
     sex: Sex
     chest_pain_type: ChestPain
-    resting_blood_pressure: int = Field(
-        ...,
-        ge=0,
-        le=250,
-        description="Resting blood pressure [mm Hg]",
-    )
-    cholesterol: int = Field(..., ge=0, le=700, description="Serum cholesterol [mm/dl]")
     fasting_blood_sugar: bool = Field(
         ...,
         description="Fasting blood sugar [1: if FastingBS > 120 mg/dl, 0: otherwise]",
     )
-    resting_electrocardiogram: RestingElectrocardiogram
     max_heart_rate: int = Field(
         ...,
         ge=60,

src/api/services/heartbeat_service.py

@@ -34,10 +34,7 @@ def create(
                 age=heartbeat_create.age,
                 sex=heartbeat_create.sex,
                 chest_pain_type=heartbeat_create.chest_pain_type,
-                resting_blood_pressure=heartbeat_create.resting_blood_pressure,
-                cholesterol=heartbeat_create.cholesterol,
                 fasting_blood_sugar=heartbeat_create.fasting_blood_sugar,
-                resting_electrocardiogram=heartbeat_create.resting_electrocardiogram,
                 max_heart_rate=heartbeat_create.max_heart_rate,
                 exercise_angina=heartbeat_create.exercise_angina,
                 old_peak=heartbeat_create.old_peak,

src/api/settings.py

@@ -1,3 +1,4 @@
+from pathlib import Path
 from pydantic import BaseModel, PostgresDsn
 from pydantic_settings import BaseSettings, SettingsConfigDict
 
@@ -15,3 +16,4 @@ class Settings(BaseSettings):
     )
 
     database: DatabaseSettings = DatabaseSettings()
+    checkpoint_path: Path