Skip to content

Hietamaki/ship-passage-predictor

Repository files navigation

Ship Movement Predictor

Uses historical AIS geodata of ship movement and builds a model based on knn method that can predict future ship movement.

PEP8 compliant, ignoring space identantion requirement (W191, E265). Also max line length is set to 99.

Module requirements

Use pip3 for packages management. You can try Anaconda if you want, but I had problems with cartopy pulling some old pyproj dependencies. If you are using pip on Windows, Gohlke's binary packages are required for uninstallable PyPi modules (https://www.lfd.uci.edu/~gohlke/pythonlibs/).

  • alphashape
  • cartopy (*)
  • descartes
  • feather-format
  • geopandas (really slow without)
  • matplotlib
  • pyepsg
  • pyproj
  • scipy
  • sklearn
  • (tables)

if you are using pip on Linux

(*) Cartopy in PyPi repository currently doesn't install. So it has to be installed from distro. With debian based:

apt install cartopy
pip3 uninstall shapely
pip3 install --no-binary shapely shapely

"The Shapely wheels on PyPI contain their own version of GEOS and you will need to avoid them by using pip's --no-binary option. This will install a shapely package which has extension modules linked against your system GEOS library." -- shapely/shapely#651 (comment)

Usage

Running for the first time downloads natural earth map data which can take up to a minute.

Multi-threading only works on Linux for now, so txt parsing will usually take 4 times as long on Windows.

About

Predicting ship's passage to target area using k-NN

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published