This project provides a set of tools for loading flight trajectory data, calculating distances between flight paths, and performing clustering using DBSCAN. The final output includes various clustering metrics such as silhouette score, Davies-Bouldin index, and more.
- Load Dataset: Load flight trajectory data from a CSV file.
- Preprocess Data: Align and normalize flight paths to ensure consistency for distance calculation.
- Calculate Distance Matrix: Compute a pairwise distance matrix between flight paths.
- DBSCAN Clustering: Perform clustering on flight trajectories using the DBSCAN algorithm.
- Clustering Metrics: Calculate various internal and external clustering evaluation metrics, including:
- Silhouette Coefficient
- Davies-Bouldin Index
- Adjusted Rand Index (ARI)
- Normalized Mutual Information (NMI)
- Final weighted score combining all metrics
- Python 3.6+
- Required libraries:
numpypandasscikit-learn
To install the necessary dependencies, run:
pip install -r requirements.txtMake sure your dataset is in a CSV format with columns x, y, z, label, and flight_id. The x, y, z columns represent the flight trajectory coordinates, label contains the true labels for evaluation, and flight_id is the unique identifier for each flight.
Use the cluster() function to perform the clustering process. Below is an example script that shows how to load the dataset and run clustering:
import pandas as pd
import clustering
# Step 1: Load the dataset
df = pd.read_csv('data/valid_data.csv')
# Step 2: Perform clustering and evaluate metrics
clustering.cluster(df, 'data/distance_df.csv')The clustering process will output:
- Progress logs for each major step.
- The final clustering metrics including the Silhouette Coefficient, Davies-Bouldin Index, ARI, NMI, and a final weighted score.
- The cluster labels for each flight trajectory.
Example output:
Begin clustering process.
Feature and label extraction complete.
Reading precomputed distance matrix...
Distance matrix loaded.
DBSCAN clustering finished.
Labels unified across flight groups.
Calculating clustering metrics...
Silhouette Coefficient: 0.6523
Davies-Bouldin Index: 0.3456
Adjusted Rand Index: 0.7802
Normalized Mutual Information: 0.8234
Final Weighted Score: 0.7765
Clustering process completed.
Cluster labels for each sample: [0, 1, 1, 0, ...]
You can modify the DBSCAN parameters directly in the perform_dbscan_clustering() function, such as:
eps: The maximum distance between two samples for one to be considered as in the neighborhood of the other.min_samples: The number of samples in a neighborhood for a point to be considered as a core point.
For example:
pre_labels = perform_dbscan_clustering(distance_matrix, eps=0.03, min_samples=10)flight_clustering/
│
├── flight_clustering.py # Main logic for clustering and metric calculation
├── requirements.txt # Required dependencies
└── README.md # Project documentation
This project is licensed under the MIT License - see the LICENSE file for details.
- slience_me - Initial work
-
Install dependencies:
- Run
pip install -r requirements.txtto install the required Python libraries.
- Run
-
Prepare your dataset:
- Ensure your data is in CSV format and contains the necessary columns (
x,y,z,label,flight_id).
- Ensure your data is in CSV format and contains the necessary columns (
-
Run the clustering:
- Use the provided example to load your dataset and execute the clustering and evaluation process.
Let me know if you need any adjustments or additional sections for the README!