'''
cd h2o-3.42.0.2
java -jar h2o.jar
http://localhost:54321
source h20env/bin/activate
jupyter notebook


'''


import h2o
import pandas as pd
import matplotlib as plt
import gradio as gr
import random
plt.use("Agg")


h2o.init()

gbm_saved_model = h2o.load_model('StackedEnsemble_AllModels_1_AutoML_1_20230812_124802')


def predict(*args):
    df = pd.DataFrame([list(args)], columns=['department','promoted','review','projects','salary','tenure','satisfaction','bonus','avg_hrs_month'])
    x = h2o.H2OFrame(df)
    pos_pred = gbm_saved_model.predict(x)
    #str(pos_pred)
    return (pos_pred.as_data_frame())




unique_department = ['IT',
 'admin',
 'engineering',
 'finance',
 'logistics',
 'marketing',
 'operations',
 'retail',
 'sales',
 'support']

unique_salary = ['high', 'low', 'medium']

with gr.Blocks() as demo:
    gr.Markdown("""
    **Employee leave probability prediction using H2O AutoML demo app**.
    Data set used - https://www.kaggle.com/datasets/marikastewart/employee-turnover .
    Jupyter Notebook is available at *Files* tab
    """)
    with gr.Row():
        with gr.Column():

            
            department = gr.Dropdown(
                label="Department",
                choices=unique_department,
                value=lambda: random.choice(unique_department),
            )

            promoted = gr.Number(

                label="Promoted",
                minimum=0.0,
                maximum=1.0
                
            )

            review = gr.Slider(label="Review", minimum=0, maximum=1, step=0.01, randomize=True)

            # review = gr.Number(
            #     label="Review",
            #     minimum=0.0,
            #     maximum=1.0
            # )

            projects = gr.Number(
                label="Projects",
                minimum=0.0,
                maximum=30.0
                
            )

            salary = gr.Dropdown(
                label="salary",
                choices=unique_salary,
                value=lambda: random.choice(unique_salary),
            )

            tenure = gr.Number(
                label="Tenure",
                minimum=0.0,
                maximum=50.0
            )

            satisfaction = gr.Slider(label="Satisfaction", minimum=0, maximum=1, step=0.01, randomize=True)

            # satisfaction = gr.Number(
            #     label="Satisfaction",
            #     minimum=0.0,
            #     maximum=1.0
            # )

            bonus = gr.Number(
                
                label="Bonus",
                minimum=0.0,
                maximum=1.0
                
            )

            avg_hrs_month = gr.Number(
                label="Avg_hrs_month",
                minimum=0.0,
                maximum=500.0
            )




        with gr.Column():
            label = gr.Dataframe()
            
            with gr.Row():
                predict_btn = gr.Button(value="Predict")
                
            predict_btn.click(
                predict,

                inputs=[
                    department,
                    promoted,
                    review,
                    projects,
                    salary,
                    tenure,
                    satisfaction,
                    bonus,
                    avg_hrs_month
                ],
                
                
                outputs=[label],
            )


demo.launch()