# app.py import gradio as gr import pandas as pd import joblib import numpy as np from sklearn.base import BaseEstimator, TransformerMixin # ------------------------------- # Load saved artifacts # ------------------------------- xgb_pipe = joblib.load("xgb_desertification_pipeline.pkl") le = joblib.load("label_encoder.pkl") expected_features = joblib.load("feature_names.pkl") # ------------------------------- # FeatureFilter class # ------------------------------- # Needed in case it's part of the pipeline class FeatureFilter(BaseEstimator, TransformerMixin): def __init__(self, expected_features): self.expected_features = expected_features def fit(self, X, y=None): return self def transform(self, X): X = X.copy() # Keep only expected features X = X[[c for c in self.expected_features if c in X.columns]] # Add missing columns with NaN for c in self.expected_features: if c not in X.columns: X[c] = np.nan return X[self.expected_features] # ------------------------------- # Prediction function # ------------------------------- def predict_desertification(**kwargs): # Convert inputs to DataFrame df = pd.DataFrame([kwargs]) df = df[expected_features] # reorder columns to match training # Predict pred_encoded = xgb_pipe.predict(df) pred_label = le.inverse_transform(pred_encoded) pred_proba = xgb_pipe.predict_proba(df).max(axis=1) return { "Desertification Level": pred_label[0], "Confidence": float(pred_proba[0]) } # ------------------------------- # Define Gradio interface # ------------------------------- # Separate categorical and numeric inputs if needed inputs = [] for feat in expected_features: if feat == "lc_type1": inputs.append(gr.Number(label=feat, value=0)) else: inputs.append(gr.Number(label=feat, value=0.0)) # Gradio app demo = gr.Interface( fn=predict_desertification, inputs=inputs, outputs="json", title="Desertification Level Predictor", description="Predict desertification level given soil, climate, and land cover features." ) # Launch demo.launch()