Abstract:This work leverages the U.S. Federal Aviation Administration's Traffic Flow Management System dataset and DV8, a recently developed tool for highly interactive visualization of air traffic data, to develop clustering algorithms for categorizing air traffic by their varying flight paths. Two clustering methodologies, a spatial-based geographic distance model, and a vector-based cosine similarity model, are demonstrated and compared for their clustering effectiveness. Examples of their applications reveal successful, realistic clustering based on automated clustering result determination and human-in-the-loop processes, with geographic distance algorithms performing better for enroute portions of flight paths and cosine similarity algorithms performing better for near-terminal operations, such as arrival paths. A point extraction technique is applied to improve computation efficiency.

What problem does this paper attempt to address?

### Problems the Paper Attempts to Solve This paper aims to address the high variability in flight paths within the United States National Airspace System. Although planned flight routes are typically predictable, actual flight paths become more random due to various factors, leading to inefficiencies in travel time, fuel consumption, and aircraft utilization. If there were a way to measure the variability between actual flight paths and planned routes, the industry could develop measures to improve efficiency for these routes. To this end, the paper proposes a flight path clustering method based on the Federal Aviation Administration (FAA) NextGen System Wide Information Management (SWIM) Traffic Flow Management System (TFMS) dataset. This method utilizes a fast interactive visualization tool called DV8 and employs two clustering distance models: a spatial model based on geographic distance and a vector model based on cosine similarity. Through these methods, the paper aims to achieve effective classification and clustering of different flight paths, thereby helping to identify and analyze patterns and anomalies in flight paths. ### Main Contributions 1. **Data Visualization and Clustering Algorithms**: The paper combines the FAA's TFMS dataset with the DV8 visualization tool to develop two clustering algorithms, based on geographic distance and cosine similarity, respectively. 2. **Comparison of Clustering Effectiveness**: Through multiple real-world cases, the paper demonstrates the performance of the two clustering methods in different scenarios, finding that the geographic distance model performs better on long-haul flight paths, while the cosine similarity model excels in terminal operations (such as arrival paths). 3. **Human-in-the-Loop Process**: The paper introduces a Human-in-the-Loop (HITL) process, allowing users to manually set clustering thresholds to obtain different clustering results, increasing the flexibility of the application. 4. **Computational Performance Optimization**: By using point extraction techniques, the paper reduces computation time while maintaining clustering performance, especially when handling large-scale datasets. ### Application Prospects This research provides effective tools and methods for the aviation industry that can be used for the analysis and optimization of flight paths, helping to improve flight efficiency, reduce fuel consumption, and enhance air traffic management. Future work will further segment flight paths, conducting clustering analysis separately for the takeoff, cruise, and landing phases to fully leverage the advantages of both models.

Categorizing Flight Paths using Data Visualization and Clustering Methodologies

Using Visualization to Improve Clustering Analysis on Heterogeneous Information Network.

Visual Analysis of Traffic Data Based on Topic Modeling (chinavis 2017).

Clustering Categorical Data Based on Distance Vectors

Visual Analysis of Group Behavior Based on Origin-Destination Data

Visualizing the Result of Fuzzy Clustering Using GIS

A Simplified Framework for Air Route Clustering Based on ADS-B Data

Flight Data Adaptive Segmentation and Classification for Fleet-level Anomaly Detection

Improved Clustering Algorithm Based on Modified Fuzzy C-means Applied to the Traffic

Clustering and Analysis of GPS Trajectory Data using Distance-based Features

A Flight Trajectory Prediction Method based on Trajectory Clustering

A trajectory clustering approach based on decision graph and data field for detecting hotspots

Clustering Of Trajectories Based On Hausdorff Distance

Low-Rank Robust Subspace Tensor Clustering for Metro Passenger Flow Modeling

A Stepwise Spatio-Temporal Flow Clustering Method for Discovering Mobility Trends

Spatial Clustering Approach for Vessel Path Identification

Modelling and application of a spectral clustering method for shared bicycle trajectories

Data Clustering in Urban Computational Modeling by Integrated Geometry and Imagery Features for Probabilistic Navigation

Clustering Activity-Travel Behavior Time Series using Topological Data Analysis

Evaluating Multi-Dimensional Visualizations for Understanding Fuzzy Clusters.