app.py

import streamlit as st
from flaml.automl.data import get_output_from_log
import plotly.express as px

from utils import csv_to_featuers_list, pre_process_df, pre_process_features
from autoML import autoML


st.set_page_config(layout="wide")

# add categorical features
# handle missing values with automl (not possible)

# fix importance features problem

with st.sidebar:

    st.subheader('Demo Datasets')
    demo_but_class = st.button(label="Demo Classification on Wine Rate Dataset")
    demo_but_reg = st.button(label="Demo Regression on California House Dataset")

    st.subheader('AutoML your Dataset')

    csv = st.file_uploader(label='CSV file')
    if csv != None:
        csv = csv.name
    
    task = st.selectbox(label='Task', options=['Classification', 'Regression'])

    if task == 'Classification':
        metric_to_minimize_class = st.selectbox(label='Metric to minimize', options=['accuracy', 'f1'])
        metric_to_minimize_reg = None
    if task == 'Regression':
        metric_to_minimize_reg = st.selectbox(label='Metric to minimize', options=['r2', 'mse', 'rmse'])
        metric_to_minimize_class = None
        
    if csv:
        lst_features = csv_to_featuers_list(csv)
        label = st.selectbox(label='Label', options=lst_features)
    
    budget = st.text_area(label='Budget Time', value="5")
    start_but = st.button(label='AutoML')


if start_but:

    autoML(csv, task, budget, label, metric_to_minimize_class, metric_to_minimize_reg)


if demo_but_class:

    autoML(csv='datasets/WineRate.csv', 
           task='Classification', 
           budget=budget, 
           label='quality', 
           metric_to_minimize_class='accuracy', 
           metric_to_minimize_reg=None)


if demo_but_reg:

    autoML(csv='datasets/house_california.csv', 
           task='Regression', 
           budget=budget, 
           label='median_house_value', 
           metric_to_minimize_class=None, 
           metric_to_minimize_reg='r2')