app.py

import gradio as gr
import numpy as np
import pickle

# Cargar el modelo
with open('model_rf.pkl', 'rb') as file:
    rf = pickle.load(file)

# Cargar el scaler
with open('my-standard-scaler.pkl', 'rb') as file:
    s_c = pickle.load(file)

# Definir la función de predicción
def predict_1(SOC: float):
    SOC = SOC/1.724 # SOC <- MO/1.724
    prediction = 1.58 + np.exp(-0.07*SOC)
    return prediction.round(2)

def predict_2(SOC: float, Cy: float):
    SOC = SOC/1.724 # SOC <- MO/1.724
    prediction = np.array(2.03 - 0.008*Cy - 0.008*SOC)
    return prediction.round(2)

def predict_3(SOC: float, Cy: float, eCa: float, eMg: float, eK: float, eNa: float, eAlH: float):
    SOC = SOC/1.724 # SOC <- MO/1.724
    ECEC = eCa + eMg + eK + eNa + eAlH
    xK = eK/ECEC
    prediction = np.array(1.53 - 0.076*SOC + 0.004*Cy - 2.04*xK)
    return prediction.round(2)
    
def predict_4(pH: float, EC: float, CCE: float, SOC: float, Sa: float, Si: float,
            Cy: float, CEC: float, eCa: float, eMg: float, eK: float, eNa: float, eAlH: float):
    SOC = SOC/1.724 # SOC <- MO/1.724
    ECEC = eCa + eMg + eK + eNa + eAlH
    xCa = eCa/ECEC
    xMg = eMg/ECEC
    xK = eK/ECEC
    xNa = eNa/ECEC
    xAlH = eAlH/ECEC
    BS1 = (eCa + eMg + eK + eNa)/CEC
    BS2 = (eCa + eMg + eK + eNa)/ECEC
    input_features = np.array([[pH,	EC,	CCE, SOC, Sa, Si, Cy, CEC, ECEC, xCa, xMg, xK, xNa, xAlH, BS1, BS2]])
    input_features_scale = s_c.transform(input_features)
    prediction = rf.predict(input_features_scale)[0].round(2)
    return prediction

# Crear la interfaz Gradio
with gr.Blocks() as demo:
    gr.Markdown("# Estimación de la densidad aparente del suelo")
    with gr.Row():
        with gr.Column():
            SOC = gr.Number(label="MO (%)", value=2.9408, interactive=True)
            eK = gr.Number(label="K (meq/100g)", value=0.47, interactive=True)
            pH = gr.Number(label="pH (1:1)", value=7.09, interactive=True)
            EC = gr.Number(label="C.E. (dS/m)", value=0.31, interactive=True)
            
        with gr.Column():
            Cy = gr.Number(label="Arcilla (%)", value=31.0, interactive=True)
            eNa = gr.Number(label="Na (meq/100g)", value=0.15, interactive=True)
            CCE = gr.Number(label="CaCO3 (%)", value=0.20, interactive=True) 
            
            
        with gr.Column():
            eCa = gr.Number(label="Ca (meq/100g)", value=19.44, interactive=True)
            eAlH = gr.Number(label="Al + H (meq/100g)", value=0.0, interactive=True)
            Si = gr.Number(label="Limo (%)", value=24.0, interactive=True)
            
        with gr.Column():
            eMg = gr.Number(label="Mg (meq/100g)", value=3.47, interactive=True)
            Sa = gr.Number(label="Arena (%)", value=45.0, interactive=True)
            CEC = gr.Number(label="CIC (meq/100g)", value=23.52, interactive=True)
            
    with gr.Row():
        with gr.Column():
            submit_1 = gr.Button(value='Abdelbaki')
        with gr.Column():
            submit_2 = gr.Button(value='Benites')
        with gr.Column():
            submit_3 = gr.Button(value='MLRegression')
        with gr.Column():
            submit_4 = gr.Button(value='Random Forest')
    output = gr.Textbox(label=" Densidad aparente del suelo (%).", interactive=False)

    submit_1.click(predict_1, inputs=[SOC], outputs=[output])
    submit_2.click(predict_2, inputs=[SOC, Cy], outputs=[output])
    submit_3.click(predict_3, inputs=[SOC, Cy, eCa, eMg, eK, eNa, eAlH], outputs=[output])
    submit_4.click(predict_4, inputs=[pH , EC, CCE, SOC, Sa, Si, Cy, CEC, eCa, eMg, eK, eNa, eAlH], outputs=[output])

demo.launch(share=False, debug=False)