""" Feedback Learning System for Agricultural Analysis Allows users to rate and provide feedback on analysis results to improve system performance """ import streamlit as st import logging from datetime import datetime, timedelta from typing import Dict, List, Any, Optional from google.cloud.firestore import FieldFilter # Use our configured Firestore client instead of direct import import json # Configure logging logger = logging.getLogger(__name__) class FeedbackLearningSystem: """Handles user feedback collection and learning system improvements""" def __init__(self): self.logger = logging.getLogger(f"{__name__}.FeedbackLearningSystem") def collect_feedback(self, analysis_id: str, user_id: str, feedback_data: Dict[str, Any]) -> bool: """ Collect user feedback for an analysis Args: analysis_id: Unique identifier for the analysis user_id: User identifier feedback_data: Dictionary containing feedback information Returns: bool: True if feedback was saved successfully """ try: db = self._get_firestore_client() if not db: self.logger.error("Failed to get Firestore client") return False # Prepare feedback document feedback_doc = { 'analysis_id': analysis_id, 'user_id': user_id, 'timestamp': datetime.now(), 'overall_rating': feedback_data.get('overall_rating', 0), 'accuracy_rating': feedback_data.get('accuracy_rating', 0), 'usefulness_rating': feedback_data.get('usefulness_rating', 0), 'clarity_rating': feedback_data.get('clarity_rating', 0), 'step_ratings': feedback_data.get('step_ratings', {}), 'recommendations_rating': feedback_data.get('recommendations_rating', 0), 'visualizations_rating': feedback_data.get('visualizations_rating', 0), 'written_feedback': feedback_data.get('written_feedback', ''), 'improvement_suggestions': feedback_data.get('improvement_suggestions', ''), 'would_recommend': feedback_data.get('would_recommend', False), 'feedback_categories': feedback_data.get('feedback_categories', []), 'session_data': { 'user_agent': feedback_data.get('user_agent', ''), 'timestamp': datetime.now().isoformat() } } # Save to Firestore feedback_ref = db.collection('user_feedback').document() feedback_ref.set(feedback_doc) self.logger.info(f"Feedback saved successfully for analysis {analysis_id}") return True except Exception as e: self.logger.error(f"Error saving feedback: {str(e)}") return False def get_feedback_analytics(self, days_back: int = 30) -> Dict[str, Any]: """ Get feedback analytics for system improvement Args: days_back: Number of days to look back for analytics Returns: Dict containing analytics data """ try: db = self._get_firestore_client() if not db: return {} # Calculate date range end_date = datetime.now() start_date = end_date - timedelta(days=days_back) # Query feedback data feedback_ref = db.collection('user_feedback') feedback_query = feedback_ref.where(filter=FieldFilter('timestamp', '>=', start_date)) feedback_docs = list(feedback_query.stream()) if not feedback_docs: return { 'total_feedback': 0, 'average_ratings': {}, 'feedback_trends': {}, 'improvement_areas': [], 'recommendation_rate': 0 } # Process feedback data total_feedback = len(feedback_docs) ratings_sum = { 'overall': 0, 'accuracy': 0, 'usefulness': 0, 'clarity': 0, 'recommendations': 0, 'visualizations': 0 } recommendation_count = 0 feedback_categories = {} improvement_suggestions = [] for doc in feedback_docs: data = doc.to_dict() # Sum ratings ratings_sum['overall'] += data.get('overall_rating', 0) ratings_sum['accuracy'] += data.get('accuracy_rating', 0) ratings_sum['usefulness'] += data.get('usefulness_rating', 0) ratings_sum['clarity'] += data.get('clarity_rating', 0) ratings_sum['recommendations'] += data.get('recommendations_rating', 0) ratings_sum['visualizations'] += data.get('visualizations_rating', 0) # Count recommendations if data.get('would_recommend', False): recommendation_count += 1 # Collect feedback categories for category in data.get('feedback_categories', []): feedback_categories[category] = feedback_categories.get(category, 0) + 1 # Collect improvement suggestions if data.get('improvement_suggestions'): improvement_suggestions.append(data['improvement_suggestions']) # Calculate averages average_ratings = { key: round(value / total_feedback, 2) if total_feedback > 0 else 0 for key, value in ratings_sum.items() } # Calculate recommendation rate recommendation_rate = round((recommendation_count / total_feedback) * 100, 2) if total_feedback > 0 else 0 # Identify improvement areas (lowest rated categories) improvement_areas = sorted( average_ratings.items(), key=lambda x: x[1] )[:3] return { 'total_feedback': total_feedback, 'average_ratings': average_ratings, 'feedback_trends': feedback_categories, 'improvement_areas': improvement_areas, 'recommendation_rate': recommendation_rate, 'improvement_suggestions': improvement_suggestions[:10] # Top 10 suggestions } except Exception as e: self.logger.error(f"Error getting feedback analytics: {str(e)}") return {} def get_learning_insights(self) -> Dict[str, Any]: """ Generate learning insights from feedback data for system improvement Returns: Dict containing learning insights and recommendations """ try: analytics = self.get_feedback_analytics(days_back=90) # 3 months of data if analytics.get('total_feedback', 0) < 5: return { 'insufficient_data': True, 'message': 'Need more feedback data to generate insights' } insights = { 'system_performance': { 'overall_score': analytics['average_ratings'].get('overall', 0), 'accuracy_score': analytics['average_ratings'].get('accuracy', 0), 'usefulness_score': analytics['average_ratings'].get('usefulness', 0), 'clarity_score': analytics['average_ratings'].get('clarity', 0) }, 'improvement_priorities': [], 'strengths': [], 'recommendations': [] } # Identify improvement priorities for area, score in analytics['improvement_areas']: if score < 3.0: # Low rating threshold insights['improvement_priorities'].append({ 'area': area, 'current_score': score, 'priority': 'high' if score < 2.5 else 'medium' }) # Identify strengths (high ratings) for area, score in analytics['average_ratings'].items(): if score >= 4.0: # High rating threshold insights['strengths'].append({ 'area': area, 'score': score }) # Generate recommendations based on feedback if analytics['average_ratings'].get('accuracy', 0) < 3.5: insights['recommendations'].append({ 'type': 'accuracy', 'suggestion': 'Improve data analysis accuracy by enhancing LLM prompts and validation' }) if analytics['average_ratings'].get('clarity', 0) < 3.5: insights['recommendations'].append({ 'type': 'clarity', 'suggestion': 'Enhance result presentation and explanation clarity' }) if analytics['average_ratings'].get('visualizations', 0) < 3.5: insights['recommendations'].append({ 'type': 'visualizations', 'suggestion': 'Improve chart and graph quality and relevance' }) return insights except Exception as e: self.logger.error(f"Error generating learning insights: {str(e)}") return {} def _get_firestore_client(self): """Get Firestore client""" try: from utils.firebase_config import get_firestore_client return get_firestore_client() except Exception as e: self.logger.error(f"Error getting Firestore client: {str(e)}") return None def display_feedback_section(analysis_id: str, user_id: str): """ Display feedback collection section in the UI Args: analysis_id: Unique identifier for the analysis user_id: User identifier """ st.markdown("---") # Feedback section header st.markdown("""

📊 Help Us Improve

Your feedback helps us enhance our agricultural analysis system

""", unsafe_allow_html=True) # Initialize feedback system feedback_system = FeedbackLearningSystem() # Create feedback form with st.form("feedback_form", clear_on_submit=True): st.markdown("### 🌟 Overall Rating") # Overall rating col1, col2, col3 = st.columns([1, 2, 1]) with col2: overall_rating = st.slider( "How would you rate this analysis overall?", min_value=1, max_value=5, value=3, help="1 = Poor, 5 = Excellent" ) st.markdown("### 📋 Detailed Ratings") # Detailed ratings in columns col1, col2 = st.columns(2) with col1: accuracy_rating = st.slider( "Accuracy of Analysis", min_value=1, max_value=5, value=3, help="How accurate were the findings and recommendations?" ) usefulness_rating = st.slider( "Usefulness of Recommendations", min_value=1, max_value=5, value=3, help="How useful were the recommendations for your farm?" ) clarity_rating = st.slider( "Clarity of Presentation", min_value=1, max_value=5, value=3, help="How clear and easy to understand were the results?" ) with col2: recommendations_rating = st.slider( "Quality of Recommendations", min_value=1, max_value=5, value=3, help="How relevant and actionable were the recommendations?" ) visualizations_rating = st.slider( "Quality of Charts/Graphs", min_value=1, max_value=5, value=3, help="How helpful were the visualizations and charts?" ) st.markdown("### 💬 Written Feedback") # Written feedback written_feedback = st.text_area( "Please share your thoughts about this analysis:", placeholder="What did you find most helpful? What could be improved?", height=100 ) # Improvement suggestions improvement_suggestions = st.text_area( "Suggestions for improvement:", placeholder="Any specific suggestions to make the analysis better?", height=80 ) # Feedback categories st.markdown("### 🏷️ Feedback Categories") feedback_categories = st.multiselect( "What aspects would you like to see improved? (Select all that apply)", [ "Data Accuracy", "Recommendation Relevance", "Chart Quality", "Explanation Clarity", "Step-by-Step Analysis", "Economic Forecasts", "Visualization Design", "Technical Details", "User Interface", "Loading Speed" ] ) # Recommendation question st.markdown("### 👍 Recommendation") would_recommend = st.radio( "Would you recommend this analysis system to other farmers?", ["Yes", "No", "Maybe"], horizontal=True ) # Submit button submitted = st.form_submit_button( "Submit Feedback", type="primary", width="stretch" ) if submitted: # Prepare feedback data feedback_data = { 'overall_rating': overall_rating, 'accuracy_rating': accuracy_rating, 'usefulness_rating': usefulness_rating, 'clarity_rating': clarity_rating, 'recommendations_rating': recommendations_rating, 'visualizations_rating': visualizations_rating, 'written_feedback': written_feedback, 'improvement_suggestions': improvement_suggestions, 'feedback_categories': feedback_categories, 'would_recommend': would_recommend == "Yes", 'user_agent': st.session_state.get('user_agent', 'Unknown') } # Save feedback if feedback_system.collect_feedback(analysis_id, user_id, feedback_data): st.success("🎉 Thank you for your feedback! Your input helps us improve the system.") st.balloons() else: st.error("❌ There was an error saving your feedback. Please try again.") def display_feedback_analytics(): """ Display feedback analytics for admin users """ st.markdown("### 📊 Feedback Analytics") feedback_system = FeedbackLearningSystem() # Get analytics analytics = feedback_system.get_feedback_analytics(days_back=30) insights = feedback_system.get_learning_insights() if analytics.get('total_feedback', 0) == 0: st.info("No feedback data available yet.") return # Display metrics col1, col2, col3, col4 = st.columns(4) with col1: st.metric( "Total Feedback", analytics['total_feedback'], help="Number of feedback submissions in the last 30 days" ) with col2: st.metric( "Overall Rating", f"{analytics['average_ratings'].get('overall', 0):.1f}/5.0", help="Average overall rating from users" ) with col3: st.metric( "Recommendation Rate", f"{analytics['recommendation_rate']:.1f}%", help="Percentage of users who would recommend the system" ) with col4: st.metric( "Accuracy Score", f"{analytics['average_ratings'].get('accuracy', 0):.1f}/5.0", help="Average accuracy rating from users" ) # Display detailed ratings st.markdown("#### 📈 Detailed Ratings") ratings_data = analytics['average_ratings'] for category, score in ratings_data.items(): if category != 'overall': st.progress(score / 5.0) st.write(f"**{category.title()}**: {score:.1f}/5.0") # Display improvement areas if insights.get('improvement_priorities'): st.markdown("#### 🎯 Improvement Priorities") for priority in insights['improvement_priorities']: priority_color = "🔴" if priority['priority'] == 'high' else "🟡" st.write(f"{priority_color} **{priority['area'].title()}**: {priority['current_score']:.1f}/5.0") # Display strengths if insights.get('strengths'): st.markdown("#### 💪 System Strengths") for strength in insights['strengths']: st.write(f"✅ **{strength['area'].title()}**: {strength['score']:.1f}/5.0") # Display recommendations if insights.get('recommendations'): st.markdown("#### 💡 Improvement Recommendations") for rec in insights['recommendations']: st.write(f"🔧 **{rec['type'].title()}**: {rec['suggestion']}")