import streamlit as st import pandas as pd import plotly.express as px import seaborn as sns import numpy as np import pickle import datetime from sklearn.model_selection import train_test_split, cross_val_score from sklearn.preprocessing import LabelEncoder, StandardScaler from sklearn.ensemble import RandomForestClassifier from sklearn.metrics import mean_squared_error, mean_absolute_error, r2_score, accuracy_score, precision_score, recall_score, f1_score, confusion_matrix import matplotlib.pyplot as plt # Load dataset and preprocess def preprocess_data(): df = pd.read_csv('fire_prediction_dataset_large.csv') label_encoder = LabelEncoder() df['Area'] = label_encoder.fit_transform(df['Area']) X = df.drop(columns=['Fire Occurrence']) y = df['Fire Occurrence'] scaler = StandardScaler() X_scaled = scaler.fit_transform(X) X_train, X_test, y_train, y_test = train_test_split(X_scaled, y, test_size=0.2, random_state=42) return X_train, X_test, y_train, y_test, df, X_scaled, y X_train, X_test, y_train, y_test, df, X_scaled, y = preprocess_data() # Train Random Forest model rf_model = RandomForestClassifier(n_estimators=100, random_state=42) rf_model.fit(X_train, y_train) y_pred = rf_model.predict(X_test) cv_scores = cross_val_score(rf_model, X_scaled, y, cv=5) # Save Random Forest model pickle.dump(rf_model, open('fire_model.pkl', 'wb')) # Load advanced model advanced_model = pickle.load(open('advanced_model.pkl', 'rb')) def predict_forest(model, oxygen, humidity, temperature): try: oxygen, humidity, temperature = float(oxygen), float(humidity), float(temperature) input_data = np.array([[oxygen, humidity, temperature]]).astype(np.float64) prediction = model.predict_proba(input_data)[0][0] return round(prediction, 2) except ValueError: return None # Streamlit App st.set_page_config(page_title="Forest Fire Prediction", page_icon="πŸ”₯", layout="wide") st.title("πŸ”₯ Forest Fire Prediction & Analysis") # Tabs for Navigation tabs = st.tabs(["πŸ“Š Model Evaluation", "πŸ”₯ Fire Risk Prediction"]) with tabs[0]: st.header("Model Performance Metrics") st.write(f'Accuracy: {accuracy_score(y_test, y_pred):.3f}') st.write(f'MSE: {mean_squared_error(y_test, y_pred):.3f}') st.write(f'MAE: {mean_absolute_error(y_test, y_pred):.3f}') st.write(f'RΒ² Score: {r2_score(y_test, y_pred):.3f}') st.write(f'Precision: {precision_score(y_test, y_pred):.3f}') st.write(f'Recall: {recall_score(y_test, y_pred):.3f}') st.write(f'F1 Score: {f1_score(y_test, y_pred):.3f}') st.subheader("Confusion Matrix") fig, ax = plt.subplots() sns.heatmap(confusion_matrix(y_test, y_pred), annot=True, fmt='d', cmap='Blues', xticklabels=['No Fire', 'Fire'], yticklabels=['No Fire', 'Fire']) plt.xlabel('Predicted') plt.ylabel('Actual') st.pyplot(fig) st.subheader("Fire Occurrence Distribution") st.plotly_chart(px.histogram(df, x='Fire Occurrence', title='Fire Occurrence Distribution', nbins=2)) st.subheader("Feature Impact") for feature in ['Temperature', 'Humidity', 'Oxygen']: st.plotly_chart(px.box(df, x='Fire Occurrence', y=feature, color='Fire Occurrence', title=f'{feature} vs Fire Occurrence')) with tabs[1]: st.header("Fire Risk Prediction") model_choice = st.selectbox("Choose Model", ["Random Forest Model", "Advanced Model"]) oxygen = st.text_input("Oxygen Level (%)", "") humidity = st.text_input("Humidity (%)", "") temperature = st.text_input("Temperature (Β°C)", "") if st.button("Predict πŸ”₯"): model = rf_model if model_choice == "Random Forest Model" else advanced_model output = predict_forest(model, oxygen, humidity, temperature) if output is None: st.error("🚨 Please enter valid numerical values.") else: st.success(f"πŸ”₯ Fire Probability: **{output * 100:.2f}%**") if output > 0.5: st.error("Forest in🚨 High Risk! of fire occurence") else: st.success("βœ… Low Risk! No worries.....") st.subheader("πŸ“œ Recent Predictions") try: with open("fire_predictions.txt", "r") as file: past_predictions = file.readlines() for pred in past_predictions[-5:]: st.text(pred.strip()) except FileNotFoundError: st.write("No previous predictions found.")