Espejo Ciudadano - Citizen Perception Dashboard

Interactive R Shiny platform for visualizing civic data from Ciudad Juárez

R Shiny

Dashboard

Survey Analytics

Data Visualization

Government

Author

Frank Leyva

Published

January 1, 2024

Executive Summary

Espejo Ciudadano (Citizen Mirror) is an interactive R Shiny dashboard developed for Así Estamos Juárez that visualizes citizen perception and participation data in Ciudad Juárez. The application presents survey results across multiple thematic areas including government, wellness, infrastructure, participation, and urban planning, allowing citizens to explore quality-of-life indicators in an intuitive and accessible manner.

🔗 Live Site: asiestamosjuarez.org

Project Context

Client Organization

Así Estamos Juárez is a civic monitoring initiative that collects and analyzes quality-of-life data in Ciudad Juárez. The organization conducts annual citizen perception and participation surveys to generate evidence that informs public dialogue and decision-making.

Problem Statement

The organization needed a digital platform that would: - Democratize access to complex civic data - Enable temporal comparisons (2023-2024) - Integrate survey data with government data sources - Provide an intuitive user experience for non-technical audiences - Scale to handle multiple years and survey types

Technical Solution

Application Architecture

I developed a modular single-page R Shiny application with the following features:

System Architecture:
├── Frontend: R Shiny + Bootstrap 5.3.1
├── Backend: R data processing
├── Database: SQLite for analytics
├── Deployment: Docker + Traefik + Let's Encrypt SSL
└── Data: Surveys + Government Data (Mexico)

Technology Stack

Core: - R Shiny: Main framework for interactive web application - tidyverse: Data manipulation and transformation - ggplot2: High-quality data visualizations - sf: Geospatial data processing - leaflet: Interactive district-level maps

Infrastructure: - Docker: Containerization for consistent deployment - Traefik: Reverse proxy with automatic SSL - Let’s Encrypt: SSL/TLS certificates - SQLite: Integrated analytics system

Data Analysis: - SPSS/Excel Import: Survey data processing - GeoJSON: Geographic data visualization - memoise: Caching for performance optimization

Key Features

1. Modular Design by Thematic Areas

The application organizes information into 5 main thematic areas:

Quality of Life

Economics, recreation, identity, environment, healthcare, education

Urban Mobility

Public transit, pedestrian movement

Institutions

Trust, accountability, representation, expectations

Infrastructure

Public services, housing, facilities

Civic Participation

Community engagement and citizen involvement

2. Data Visualization System

Implemented Visualization Types: - Comparative bar charts by district - Time series for trend analysis - Interactive choropleth maps with Leaflet - Indicator dashboards with key metrics - Frequency distributions for categorical data

Example visualization code:

# Automatic question classification system
question_classifier.R detects question type
→ Routes to appropriate module (bar_chart, map, time_series)
→ Applies consistent visual theme
→ Pre-caches results for performance

3. Multi-Source Data Integration

The application integrates: - Survey Data: Citizen perception (PER_2023, PER_2024) - Participation Data: Civic engagement (PAR_2023, PAR_2024) - Government Data: Official statistics from Mexico - Geographic Data: District boundaries in GeoJSON format

4. Integrated Analytics System

I developed a custom analytics system to track usage:

AnalyticsManager Class:
├── Session tracking with duration analysis
├── Section engagement metrics
├── User interaction logging
├── Daily usage statistics
└── Data export capabilities

Analytics Triggers: - Keyboard: Ctrl+Shift+A - UI: Triple-click “Vista Rápida” title - Console: showAnalytics() function

5. Performance Optimization

Implemented Strategies: - Extensive caching: Use of memoise for expensive operations - Pre-computation: Plots and percentages stored as .rds - Background loading: Asynchronous data pre-loading - Efficient data management: data_manager.R optimizes loading - Multi-core processing: Limited to 4 cores in Docker

Project Visualizations

Interactive maps — Interactive map visualizations

District analysis — District-level analysis

Technical Challenges and Solutions

Challenge 1: Scalable Modular Architecture

Problem: Need to add new themes and years without major refactoring

Solution: Implemented modular system with *_ui.R / *_server.R pairs:

R/
├── wellness/wellness_ui.R + wellness_server.R
├── government/government_ui.R + government_server.R
└── [New modules follow same pattern]

Challenge 2: Multi-Survey Configuration

Problem: Different surveys use different column names and encodings

Solution: Created survey_config.R with centralized mappings:

survey_config.R:
├── Column mappings (district, gender, age)
├── Categorical value mappings
└── Survey-specific configurations

Challenge 3: Performance with Large Datasets

Problem: Slow loading times with real-time calculations

Solution: 3-stage processing pipeline:

1. Pre-processing → data/processed/ (CSV)
2. Pre-computation → data/plots/ (.rds)
3. In-memory caching → memoise + data_manager

Challenge 4: Secure Production Deployment

Problem: Need for HTTPS, SSL, and monitoring

Solution: Complete Docker stack:

docker-compose.yml:
├── Shiny App Container (Port 3838)
├── Traefik Reverse Proxy
├── Let's Encrypt automatic SSL
└── Security header configuration

Data Pipeline

graph LR
A[Raw Data<br/>Excel/SPSS] --> B[Cleaning<br/>CSV]
B --> C[Classification<br/>question_classifier]
C --> D[Pre-computed<br/>Visualizations]
D --> E[Presentation<br/>UI Modules]

Raw Data: Excel/SPSS files in data/raw/
Processing: Cleaning and classification
Processed Data: Clean CSVs in data/processed/
Visualization: Pre-computed plots stored as .rds
Presentation: Modular UI components present the data

Code Examples

Automatic Question Classification

# question_classifier.R
classify_question <- function(question_data) {
  if (is_geographic(question_data)) {
    return(list(type = "map", module = "map_module"))
  } else if (is_temporal(question_data)) {
    return(list(type = "timeseries", module = "timeseries_module"))
  } else if (is_categorical(question_data)) {
    return(list(type = "bar", module = "bar_chart_module"))
  }
}

Analytics System

# analytics.R
AnalyticsManager <- R6Class("AnalyticsManager",
  public = list(
    log_interaction = function(session_id, section, action) {
      dbWriteTable(conn, "interactions", data.frame(
        session_id = session_id,
        section = section,
        action = action,
        timestamp = Sys.time()
      ), append = TRUE)
    },

    get_section_metrics = function() {
      dbGetQuery(conn, "
        SELECT section, COUNT(*) as visits,
               AVG(duration) as avg_duration
        FROM sessions
        GROUP BY section
      ")
    }
  )
)

Results and Impact

✅ Successful Deployment: Live site at asiestamosjuarez.org ✅ User Experience: Intuitive interface accessible to non-technical audiences ✅ Performance: Fast loading times with pre-computed data and caching ✅ Scalability: Modular architecture supports multiple years and survey types ✅ Security: HTTPS with automatic SSL via Let’s Encrypt ✅ Analytics: Tracking system provides usage insights

Performance Metrics: - Initial load time: < 3 seconds - Visualization interaction: Instantaneous (pre-cached) - Multi-device support: Desktop, tablet, mobile - Availability: 99.9% uptime with Docker

Future Work and Potential Improvements

Add data export capabilities
Implement advanced multi-year comparisons
Integrate more government data sources
Add predictive analytics and trends
Develop REST API for programmatic access
Implement admin dashboard

Technical Skills Demonstrated

This project showcases proficiency in:

R Shiny Development Interactive web applications, modular design, performance optimization

Data Science Survey analysis, data visualization, multi-source integration

DevOps & Deployment Docker, Traefik, SSL/TLS, CI/CD, infrastructure management

Geospatial Analysis GeoJSON processing, Leaflet maps, district-level analysis

Software Architecture Modular design, state management, separation of concerns

Database Systems SQLite, schema design, optimized queries, analytics

Conclusion

The Espejo Ciudadano project represents my ability to develop real-world data applications that serve the public interest. By combining survey data analysis, government data integration, interactive visualization, and modern deployment, I created a platform that empowers citizens with accessible and actionable information.

This project demonstrates not only technical competence in R Shiny and data science, but also the ability to translate complex stakeholder requirements into robust, scalable, and user-centric software solutions.

Technologies: R Shiny, Docker, Traefik, ggplot2, leaflet, SQLite, Bootstrap, GeoJSON Repository: Private (client) Live Site: asiestamosjuarez.org Client: Así Estamos Juárez - Civic Monitoring Initiative