main.py

from fastapi import FastAPI, Query, Request, HTTPException
from fastapi.responses import JSONResponse, HTMLResponse
from fastapi.templating import Jinja2Templates
import joblib
import pandas as pd
import numpy as np
from pydantic import BaseModel
from sklearn.pipeline import Pipeline
from sklearn.feature_selection import SelectKBest
from sklearn.ensemble import BaggingClassifier
from xgboost import XGBClassifier
from sklearn.preprocessing import StandardScaler
from sklearn.compose import ColumnTransformer
from sklearn.feature_selection import f_classif
from sklearn.impute import SimpleImputer 



app = FastAPI()
templates = Jinja2Templates(directory="templates")

class InputFeatures(BaseModel):
    Plasma_glucose: float
    Blood_Work_Result_1: float
    Blood_Pressure: float
    Blood_Work_Result_2: float
    Blood_Work_Result_3: float
    Body_mass_index: float
    Blood_Work_Result_4: float
    patients_age: int

# Load the pickled XGBoost model
model_input = joblib.load("model_1.joblib")

@app.post("/sepsis_prediction")
async def predict(input: InputFeatures):  

    # { Plasma_glucose: float,
    # Blood_Work_Result_1: float,
    # Blood_Pressure: float,
    # Blood_Work_Result_2: float,
    # Blood_Work_Result_3: float,
    # Body_mass_index: float,
    # Blood_Work_Result_4: float,
    # patients_age: int
    # }
    # return input
    # Numeric Features 
    num_attr = [
        ['Plasma_glucose', 'Blood_Work_Result_1', 'Blood_Pressure',
         'Blood_Work_Result_2', 'Blood_Work_Result_3', 'Body_mass_index',
         'Blood_Work_Result_4', 'patients_age']]

    #creating pipelines
    num_pipeline= Pipeline([('imputer', SimpleImputer()),('scaler', StandardScaler())])

    
    full_pipeline=ColumnTransformer([('num_pipe',num_pipeline,num_attr)])
    df = pd.DataFrame([input])
    full_pipeline.fit(df)  # Fit the full pipeline on the DataFrame

    final_input = np.array(full_pipeline.transform(df), dtype=np.str)
    prediction = model_input.predict(final_input)

    XGB = Pipeline([
        ("col_trans", full_pipeline),  # You need to define full_pipeline
        ("feature_selection", SelectKBest(score_func=f_classif, k='all')),
        ("model", BaggingClassifier(base_estimator=XGBClassifier(random_state=42)))
    ])

    # df = pd.DataFrame([input])
    # final_input = np.array(full_pipeline.fit_transform(df), dtype=np.str)  # Check predict_input, maybe it should be XGB
    # #prediction = model_input.predict(np.array([final_input]).reshape(1, -1))
    # prediction = model_input.predict(final_input)

    # full_pipeline = ColumnTransformer([('num_pipe', num_pipeline, num_attr)])
    #     full_pipeline.fit(df)  # Fit the full pipeline on the DataFrame

    #     #final_input = np.array(full_pipeline.transform(df), dtype=np.str)

    #print(model_input)
    #df = pd.DataFrame([input])
    #final_input = np.array(predict_input.fit_transform(df), dtype = np.str)
        


    return prediction

if __name__ == '__main__':
     uvicorn.run("Main:app", reload=True)





# @app.get("/")
# async def read_root():
#     return {"message": "Welcome to the Sepsis Prediction API."}

# @app.get("/form/")
# async def show_form():

# @app.post("/predict/")
# async def predict_sepsis(
#     request: Request,
#     input_data: InputFeatures  # Use the Pydantic model for input validation
# ):
#     try:
#         # Convert Pydantic model to a DataFrame for prediction
#         input_df = pd.DataFrame([input_data.dict()])

#         # Make predictions using the loaded XGBoost model
#         prediction = xgb_model.predict_proba(xgb.DMatrix(input_df))

#         # Create a JSON response
#         response = {
#             "input_features": input_data,
#             "prediction": {
#                 "class_0_probability": prediction[0],
#                 "class_1_probability": prediction[1]
#             }
#         }

#         return templates.TemplateResponse(
#             "display_params.html",
#             {
#                 "request": request,
#                 "input_features": response["input_features"],
#                 "prediction": response["prediction"]
#             }
#         )
#     except Exception as e:
#         #raise HTTPException(status_code=500, detail="An error occurred while processing the request.")
#         raise HTTPException(status_code=500, detail=str(e))