import gradio as gr

import pickle
import sklearn
import pandas as pd
import numpy as np
import joblib
from sklearn.neural_network import MLPRegressor
from sklearn.preprocessing import StandardScaler

filename= './gdp_model.sav'
scalerfile= './gdp_scale.sav'
model = pickle.load(open(filename, 'rb'))
scaler = pickle.load(open(scalerfile, 'rb'))

example_niti = pd.read_csv('./NITIDATA.csv')
example_x = np.array(example_niti.iloc[:, 0:])
example_y = example_niti.iloc[:, -1]

def predict(Capital_Receipts,
        Aggregate_Receipts,
        Social_Sector_Expenditure,
        Interest_Payments,
        Own_Tax_Revenues,
        Fiscal_Deficits,
        Outstanding_Liabilities,
        Aggregate_Expenditure,
        Revenue_Receipts,
        Revenue_Expenditure,
        Revenue_Deficits,
        Capital_Expenditure,
        expected_GDP_Only_for_examples):
    

    X = scaler.transform([[Capital_Receipts,
                            Aggregate_Receipts,
                            Social_Sector_Expenditure,
                            Interest_Payments,
                            Own_Tax_Revenues,
                            Fiscal_Deficits,
                            Outstanding_Liabilities,
                            Aggregate_Expenditure,
                            Revenue_Receipts,
                            Revenue_Expenditure,
                            Revenue_Deficits,
                            Capital_Expenditure]])
    
    return np.round(model.predict(X), 2)


demo = gr.Interface(
    predict,
    [
        "number",
        "number",
        "number",
        "number",
        "number",
        "number",
        "number",
        "number",
        "number",
        "number",
        "number",
        "number",
        "number"
    ],
    "number",
    examples=[
        list(element) for element in example_x
    ],
).launch(share=True);