Interactive Dashboard

1 Overview

This page documents the interactive Panel dashboard developed to explore flight diversion patterns. The dashboard allows users to filter and analyze diversions in real-time using multiple interactive controls.

2 Dashboard Features

2.1 Interactive Filters

The dashboard includes three main filters that work together to subset the data:

1. Airline Selection - Multi-select dropdown to choose one or more airlines - Default: Top 3 airlines by diversion count - Enables airline-specific analysis and comparison

2. Date Range Slider - Specify custom date ranges (July 2021 - December 2024) - Allows temporal analysis of diversions - Supports seasonal pattern exploration

3. Dynamic Summary Statistics - Updates automatically when filters change - Displays: - Total diversions in filtered dataset - Number of unique diversion airports - Average departure delay (in minutes)

2.2 Visualizations

2.2.1 Top Diversion Airports Map

  • Interactive geographic map showing the top 15 diversion airports
  • Marker size represents frequency of diversions
  • Hover over airports to see exact diversion counts
  • Allows geographic pattern identification

What it shows: Which airports are receiving the most diverted flights during the selected time period and for chosen airlines.

How to interpret: Larger circles = more diversions. Geographic clusters suggest regional capacity issues.

2.2.2 Diversions by Airline Bar Chart

  • Vertical bar chart ranking airlines by diversion count
  • Colors distinguish between airlines
  • Hover for exact numbers

What it shows: How diversions are distributed across airlines in your filtered dataset.

How to interpret: Comparison of operational disruption frequency by carrier. May reflect differences in route networks, fleet size, or operational practices.

2.2.3 Summary Statistics Cards

  • Three key metrics displayed prominently
  • Background colors for quick visual scanning
  • Update in real-time as filters change

Metrics: 1. Total Diversions (Blue) - Sum of all diversions matching filters 2. Diversion Airports (Purple) - Count of unique airports receiving diversions 3. Avg Departure Delay (Green) - Mean delay for diverted flights

3 How to Use the Dashboard

3.1 Running Locally

To run the dashboard on your computer:

# Navigate to project directory
cd MUSA5500-finalProject

# Activate geospatial environment
conda activate geospatial

# Run the app
python app.py

Then open your browser and go to: http://localhost:5006

3.2 Basic Workflow

  1. Start with full dataset
    • All airlines selected
    • Full date range (July 2021 - December 2024)
  2. Select airlines of interest
    • Click on airline dropdown
    • Choose one or more carriers
    • Dashboard updates automatically
  3. Narrow date range
    • Use date slider to focus on time period
    • Helpful for identifying seasonal patterns
    • Can zoom into specific disruption events
  4. Analyze results
    • Check summary statistics
    • Explore geographic distribution on map
    • Compare airlines using bar chart

3.3 Example Analyses

Find Peak Diversion Periods - Set date range to one month - Select all airlines - Identify which airports spike in activity

Compare Airline Performance - Set date range to full dataset - Select one airline at a time - Compare summary statistics across carriers

Explore Regional Events - Focus on date range when you know a disruption occurred - Select relevant airlines - Analyze geographic concentration of diversions

4 Technical Details

4.1 Technology Stack

  • Framework: Panel (by HoloViz)
  • Data Processing: Pandas
  • Visualization: Plotly
  • Data Source: BTS OTMC-OTP database (2021-2024)

4.2 Cross-Filtering Implementation

The dashboard uses Panel’s reactive decorators (@pn.depends()) to automatically update all visualizations when any filter changes. This means: - Changing airline selection updates the map and chart - Adjusting date range recalculates all statistics - All changes are instantaneous (no need to click “Update”)

4.3 Performance

  • Dashboard loads 8,651 total diversion records
  • Filtering operations complete in <1 second
  • Map rendering optimized for smooth interaction

5 Insights from Dashboard

5.1 Key Patterns Observable

Geographic Concentration The map clearly shows which airports serve as primary diversion hubs. Common patterns include: - West Coast: ONT, PHX, LAX, SFO - Texas: IAH, DFW, SAT, AUS - Midwest: IND, STL, MKE - Southeast: ATL, MIA, MCO

Airline Variation Significant differences exist in diversion frequency by airline, driven by: - Route network size and destinations served - Fleet composition and aircraft types - Operational practices and disruption management - Timing of analysis period (some airlines may have more flights during studied period)

Temporal Patterns Using the date range filter reveals: - Peak diversion activity in winter months (December) - Regional variations in seasonal patterns - Major disruption events concentrated around specific dates

5.2 Using Dashboard for Decision-Making

For Airlines: - Identify routes with highest diversion risk - Plan capacity and staffing for vulnerable airports - Compare performance to competitors

For Airports: - Understand diversion traffic patterns - Plan for overflow capacity during peak periods - Coordinate with other regional airports

For Researchers: - Explore relationships between variables - Generate hypotheses about disruption causes - Validate findings from other analyses

6 Limitations

  • No causal information: Dashboard shows patterns but not causes
  • Historical data only: Cannot predict future diversions
  • Aggregated view: Individual flight details not shown
  • Weather context missing: No concurrent weather data available
  • First diversion only: Secondary diversions not tracked

7 Future Enhancements

Potential improvements to dashboard:

  1. Add weather overlay - Show weather conditions during diversion periods
  2. Animated time series - Play through months/years to see patterns evolve
  3. Route-specific analysis - Filter by origin and destination airports
  4. Delay correlation - Show relationship between diversion frequency and delays
  5. Predictive model - Use ML to predict diversion likelihood for future flights
  6. Download data - Export filtered dataset as CSV

8 Questions & Support

For questions about: - How to use: See “How to Use” section above - Data interpretation: See “Insights from Dashboard” section - Technical issues: Make sure you have Panel, Plotly, and Pandas installed in your geospatial environment - Analysis questions: See full documentation at main site

Dashboard Created: December 2024 Data Current Through: December 2024