Skip to content

Latest commit

 

History

History
34 lines (23 loc) · 1.43 KB

README.md

File metadata and controls

34 lines (23 loc) · 1.43 KB

heatmap_expts

This project also depends on:

You will need to add the 'python' subdirectory of each project to the Python path.

The experiments in the paper can be run as follows. For synthetic data:

  • synthetic_experiments/run_synthetic_noise.py
  • synthetic_experiments/run_synthetic_noise_nogrid.py
  • synthetic_experiments/run_synthetic_bias.py

Then run plot_performance.py to plot the performance metrics. These will be saved in the output/synth folder.

For the PRN satellite image dataset:

  • prn_experiments/prn_simulation
  • plot_performance.py produces images in output/prn

For Ushahidi dataset:

  • ushahidi_experiments/ushahidi_loader_emergencies.py

For plotting heatmaps, from the repository HeatMapBCC run python/crowdscanner/ushahidiheatmap.py

The class that implements heatmapBCC is in the repository https://github.com/edwinrobots/HeatMapBCC in file python/heatmapbcc.py. The main functions you need are:

  • constructor for passing in hyperparameters
  • combine_classifications() to train the model and predict at the observed report locations
  • predict() to predict at new locations after you have trained the model using combined_classifications()