added app.py and models

app.py

@@ -0,0 +1,96 @@
+import pickle
+
+import gradio as gr
+import joblib
+import numpy as np
+import pandas as pd
+from tensorflow.keras.models import load_model
+
+# Load models
+log_reg = joblib.load("models/logistic_regression_model.pkl")
+xgb = pickle.load(open("models/xgboost_model.pkl", "rb"))
+ffnn = load_model("models/ffnn_model.keras")
+scaler = joblib.load("models/standard_scaler.pkl")
+
+import json
+
+with open("data/feature_names.json", "r") as f:
+    feature_names = json.load(f)
+
+
+def predict_default(*inputs):
+    processed_inputs = []
+    for name, val in zip(feature_names, inputs):
+        if name in categorical_mappings:
+            val = categorical_mappings[name].index(val)  # Convert string to int
+        processed_inputs.append(val)
+
+    input_df = pd.DataFrame([processed_inputs], columns=feature_names)
+    scaled = scaler.transform(input_df)
+
+    logit = log_reg.predict_proba(scaled)[0][1]
+    xgb_pred = xgb.predict_proba(input_df.values)[0][1]
+    ffnn_pred = ffnn.predict(scaled)[0][0]
+
+    return {
+        "Logistic Regression": float(logit),
+        "XGBoost": float(xgb_pred),
+        "FFNN": float(ffnn_pred),
+    }
+
+
+default_values = [
+    56.0,  # Age
+    85994.0,  # Income
+    50587.0,  # LoanAmount
+    520.0,  # CreditScore
+    80.0,  # MonthsEmployed
+    4.0,  # NumCreditLines
+    15.23,  # InterestRate
+    36.0,  # LoanTerm
+    0.44,  # DTIRatio
+    0.0,  # Education
+    0.0,  # EmploymentType
+    0.0,  # MaritalStatus
+    1.0,  # HasMortgage
+    1.0,  # HasDependents
+    4.0,  # LoanPurpose
+    1.0,  # HasCoSigner
+    -0.895272,  # AffRatio
+    0.431883,  # TotalInterest
+    0.139637,  # Debt
+    -1.28165,  # AvgBorrowed
+]
+
+categorical_mappings = {
+    "Education": ["Bachelor's", "High School", "Master's", "PhD"],
+    "EmploymentType": ["Full-time", "Part-time", "Self-employed", "Unemployed"],
+    "MaritalStatus": ["Divorced", "Married", "Single"],
+    "HasMortgage": ["No", "Yes"],
+    "HasDependents": ["No", "Yes"],
+    "LoanPurpose": ["Auto", "Business", "Education", "Home", "Other"],
+    "HasCoSigner": ["No", "Yes"],
+}
+
+
+input_components = []
+for name, val in zip(feature_names, default_values):
+    if name in categorical_mappings:
+        choices = categorical_mappings[name]
+        input_components.append(
+            gr.Dropdown(label=name, choices=choices, value=choices[int(val)])
+        )
+    else:
+        input_components.append(gr.Number(label=name, value=val))
+output_components = gr.Label(num_top_classes=3)
+
+demo = gr.Interface(
+    fn=predict_default,
+    inputs=input_components,
+    outputs=output_components,
+    title="Loan Default Risk Predictor",
+    description="Enter borrower info and see the default risk prediction from 3 models.",
+    flagging_mode="never",
+)
+
+demo.launch()

data/feature_names.json

@@ -0,0 +1 @@
+["Age", "Income", "LoanAmount", "CreditScore", "MonthsEmployed", "NumCreditLines", "InterestRate", "LoanTerm", "DTIRatio", "Education", "EmploymentType", "MaritalStatus", "HasMortgage", "HasDependents", "LoanPurpose", "HasCoSigner", "AffRatio", "TotalInterest", "Debt", "AvgBorrowed"]

models/logistic_regression_model.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:95373bda426cb336224257d4a4fa794c73403e9f1faa9e8c731e642003f44c9e
+size 1051

models/standard_scaler.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f8845ddfdfee4999895437aa2d46f98ade95d5b11ba637a5c581940c6361a159
+size 1607

models/xgboost_model.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d546ba56110501c951797c7758c516ab7f4523e2ee43778f1590eb081155de89
+size 177107

requirements.txt

@@ -0,0 +1,5 @@
+gradio==5.41.0
+joblib==1.5.1
+numpy==2.3.2
+pandas==2.3.1
+tensorflow==2.20.0rc0