Data and code for the paper Detecting Road Network Errors from Trajectory Data with Partial Map Matching and Bidirectional Recurrent Neural Network Model

<pre>This repository contains data and codes for the manuscript _Detecting Road Network Errors from Trajectory Data with Partial Map Matching and Bidirectional Recurrent Neural Network Model_ submitted to IJGIS. <br><br>Requirements (tested with Python 3.8.8):<br><br>- Numpy &gt;= 1.22.4<br>- PyTorch &gt;= 2.0.0<br>- Sklearn &gt;= 0.24.1 <br>- Shapely &gt;= 1.8.2<br>- rtree &gt;= 1.0.0<br>- Networkx &gt;= 2.8.4<br>- Geopandas &gt;= 0.10.2<br><br>### Code structure <br><br>- lib: codes for partial map matching, context feature extraction and BiRNN model<br>- data: a sample of training and test dataset in npz format, which contains "xy" and "label" attribute storing the trajectory geometry<br>and manually labels. It also contains a shapefile of road network. <br><br>### Run the program with two steps <br><br>```bash<br># 1. Generate context features<br><br>python feature_extract.py<br><br># 2. Training a BiRNN model for classification<br><br>python train_birnn.py<br>```<br><br>For demonstration, data folder contains a small trajectory dataset with labels and road network downloaded from OSM. Accessing the complete dataset can be applied at https://outreach.didichuxing.com/. </pre><br>