Trajectories are one way to think about how things move, from medical equipment within hospitals to taxis in cities to planes across the globe. Although each of these things move very differently in scale and pattern, their movements can all be represented as a series of x,y points.
For our masters capstone as Data Science students at Columbia University, my team and I worked with a mentor from GE Research to cluster those series of points, those trajectories, by shape, because before deciding how things should move, to minimize costs of transport, cost of inventory, etc. we need to understand how they do move. This repository contains a selection of our work.
Autoencoders: Neural network autoencoder results
LSTM_model_shifted_Tabitha.ipynb: Compares results of training autoencoders to predict the original trajectory versus the trajectory shifted forward by oneEvaluationDataset_Tabitha.ipynb: Results of best performing model (as trained on simulated data) on another dataset
DistanceMetrics: Clustering algorithms and distance metrics designed for spaciotemporal data
clustereditdist.py: Functions for distance metrics and clustering algorithm (dbscan)EditDistance.py: Functions for edit distance algorithm, designed for spaciotemporal dataClusteringTrajectoriesDemo.ipynb: Demonstration of clustering data with dbscanClustering Trajectories with Edit Distance.ipynb: Results of clustering simulated datasets with edit distance as distance metrics and dbscan as clustering algorithm
Visualization: Visualization using LineCollection
visualizationclusters.py: Script for visualization using python LineCollection library
GE Asset Tracking_Final Report.pdf: Report describing project concept, development process, and results