Built on this repository. This network of Manhattan consists of 4,091 nodes and 9,452 edges. The travel times on each edge (road segment) during each hour of the day are provided. Taxi trip data on several days is also provided.

Download code and data files from the releases. Data files should be located in the root directory of the code.

|-- Manhattan-Map
   |-- map-data-gitignore
   |-- pickle-files-gitignore
   |-- precomputed-tables-gitignore
   |-- taxi-data-gitignore

Download the the taxi trip data file from TLC Trip Record Data, rename the data file name to what you like (e.g. yellow_tripdata_2015-05.csv) and put it in the root folder (do not forget to check the file name used in code), and run run.py, which will do the following things.

1. Run network_generator.py to load map data, build the Manhattan network using networkx and save it as a pickle file.
2. Use path_table_generator.py to precompute the shortest paths among all node pairs, store the paths and the means (and variances) of the paths in tables.
3. Use the function store_network_to_pickle() and store_path_table_to_pickle() in data_serializer.py to save these files to picklefiles, so that we can save time on loading them.
4. Run trip_filter to filter out the trip data on selected day.

Uber Movement provide travel speed (consisting of mean and std speed values on each road segment) data and other interesting data. But this code repository has not successfuly made it to process Uber Movement data.