Update app.py

Author: eklavyaxx

app.py

@@ -1,5 +1,6 @@
 import streamlit as st
 import pandas as pd
+import plotly.graph_objects as go
 from Archives.data_processing import clean_data, split_data
 from Archives.model_training import get_models, train_and_evaluate_models
 from Archives.visualisation import plot_model_comparison, plot_confusion_matrices
@@ -15,13 +16,14 @@
     text_color = "#FFFFFF"
     button_color = "#444"
     success_text_color = "#FFFFFF"
+    plotly_template = "plotly_dark"
 else:
     bg_color = "#F8F4E1"
     text_color = "#000000"
     button_color = "#333333"
     success_text_color = "#000000"
+    plotly_template = "plotly_white"
 
-# Custom CSS
 st.markdown(f"""
     <style>
         .stApp {{
@@ -43,19 +45,20 @@
         div[data-testid="stAlertContainer"] p {{
             color: {success_text_color} !important;
         }}
+        h2, h3, h4, h5, h6, .stMarkdown {{
+            color: {text_color} !important;
+        }}
+        .css-10trblm, .stMetric label, .stMetric div {{
+            color: {text_color} !important;
+        }}
     </style>
 """, unsafe_allow_html=True)
 
-# Sidebar title
 st.sidebar.title("📁 Upload and Train")
 
-# File upload
 uploaded_file = st.sidebar.file_uploader("Upload a preprocessed dataset (CSV)", type=["csv"])
-
-# Default path
 default_path = "Loan-Creaditworthiness-classification-main/data/Preprocessed/final.csv"
 
-# Load dataset
 if uploaded_file:
     df = pd.read_csv(uploaded_file)
     st.sidebar.success("✅ Custom dataset loaded successfully!")
@@ -67,19 +70,30 @@
         st.sidebar.error(f"Failed to load default dataset: {e}")
         df = None
 
-# Training and output
-if df is not None and st.sidebar.button("🚀 Train Models"):
+use_best_model = st.sidebar.checkbox("🚀 Use Best Model Automatically")
+
+if df is not None and st.sidebar.button("⚙️ Train Models"):
     with st.spinner("Training in progress... Please wait ⏳"):
         df_cleaned = clean_data(df)
         X_train, X_test, y_train, y_test = split_data(df_cleaned, target_column='high_risk_applicant')
         models = get_models()
         results, predictions = train_and_evaluate_models(models, X_train, X_test, y_train, y_test)
+
         figs = plot_model_comparison(results)
         confusion_figs = plot_confusion_matrices(predictions)
 
+        model_scores = {name: score['Accuracy'] for name, score in results.items()}
+        model_dict = {name: predictions[name][1] for name in predictions}
+        best_model_name = max(model_scores, key=model_scores.get)
+        best_model = model_dict[best_model_name]
+
+        selected_model_name = best_model_name
+        selected_model = best_model
+        st.success(f"✅ Best model auto-selected: **{best_model_name}** (Accuracy: **{model_scores[best_model_name]:.2f}**)")
+
     st.success("🎉 Training completed!")
 
-    st.subheader("📊 Model Performance")
+    st.subheader("📊 Model Performance (Bar Graphs)")
     for i in range(0, len(figs), 2):
         cols = st.columns(2)
         for j, fig in enumerate(figs[i:i+2]):
@@ -90,3 +104,36 @@
         cols = st.columns(2)
         for j, fig in enumerate(confusion_figs[i:i+2]):
             cols[j].pyplot(fig)
+
+    st.subheader("📊 Visual Comparison of All Models")
+    metrics = ['Accuracy', 'Precision', 'Recall', 'F1 Score']
+    data = {metric: [results[model].get(metric, 0) for model in results] for metric in metrics}
+    model_names = list(results.keys())
+
+    fig = go.Figure()
+    for metric in metrics:
+        fig.add_trace(go.Bar(
+            y=model_names,
+            x=data[metric],
+            name=metric,
+            orientation='h'
+        ))
+
+    fig.update_layout(
+        barmode='group',
+        title="Model Performance Comparison",
+        xaxis_title="Score",
+        yaxis_title="Models",
+        template=plotly_template,
+        height=500
+    )
+
+    st.plotly_chart(fig, use_container_width=True)
+
+    st.markdown(f"<h3 style='color:{text_color};'>📈 Real-Time Performance Comparison</h3>", unsafe_allow_html=True)
+    for name, score_dict in results.items():
+        cols = st.columns(4)
+        cols[0].markdown(f"<div style='color:{text_color}; font-weight:bold;'>{name}</div>", unsafe_allow_html=True)
+        cols[1].markdown(f"<div style='color:{text_color};'>✅ Accuracy<br><span style='font-size:28px;'><b>{score_dict['Accuracy']:.3f}</b></span></div>", unsafe_allow_html=True)
+        cols[2].markdown(f"<div style='color:{text_color};'>🎯 Precision<br><span style='font-size:28px;'><b>{score_dict['Precision']:.3f}</b></span></div>", unsafe_allow_html=True)
+        cols[3].markdown(f"<div style='color:{text_color};'>📊 Recall<br><span style='font-size:28px;'><b>{score_dict['Recall']:.3f}</b></span></div>", unsafe_allow_html=True)