app.py

from fastapitableau import FastAPITableau
from joblib import load
import pandas as pd
from typing import List

# Load model
model = load("model.joblib")

# Define model_pipline
def model_pipeline(dict):
    model_data = pd.DataFrame(dict)
    model_data["ship_diff"] = model_data.days_to_ship_actual - model_data.days_to_ship_scheduled
    pred_columns = [
        "ship_diff",
        "quantity",
        "sales",
        "discount"
    ]
    return model.predict(model_data.loc[:, pred_columns]).tolist()

# Define the extension
app = FastAPITableau(
    title = "Predicted Profit",
    description = "A simple linear prediction of sales profit given new input data",
    version = "0.1.0"
)

@app.post("/predict")
async def predict(
        days_to_ship_actual: List[int],
        days_to_ship_scheduled: List[int],
        quantity: List[int],
        sales: List[float],
        discount: List[float]
    )->List[float]:
        data = {
            "days_to_ship_actual": days_to_ship_actual,
            "days_to_ship_scheduled": days_to_ship_scheduled,
            "quantity": quantity,
            "sales": sales,
            "discount": discount
        }
        return model_pipeline(data)