Skip to content

Latest commit

 

History

History
68 lines (53 loc) · 3.25 KB

README.md

File metadata and controls

68 lines (53 loc) · 3.25 KB

DeepMove

Requirements

  • python 3.8.5
  • tensorflow 2.4.1
  • numpy 1.19.5
  • json 2.0.9
  • pickle 4.0
  • matplotlib 3.4.2

Structure

already in DeepMove dir

  • /codes

    • main_tf.py
    • model_tf.py # define models
    • train_tf.py # define tools for train the model

  • /data # preprocessed foursquare sample data (pickle file)

  • /results # the default save path when training the mode

Usage

The codes contain four network model (simple, simple_long, attn_avg_long_user, attn_local_long) and a baseline model (Markov). The parameter settings for these model can refer to result.md file.

  • Train a new model:

    python3 main_tf.py --model_mode=attn_local_long --pretrain=0

    Other parameters (refer to main_tf.py):

    • for training:
      • learning_rate, lr_step, lr_decay, L2, clip, epoch_max, dropout_p
    • model definition:
      • loc_emb_size, uid_emb_size, tim_emb_size, hidden_size, rnn_type, attn_type
      • history_mode: avg, avg, whole

  • Plot user's predicted and real trajectory:

    python3 main_tf.py --model_mode=xxx --plot_user_traj=user # user is valid from 0 to 885

    Then in each epoch, the user's trajtectory is saved as a png in /codes dir. The png would update automatically after each epoch.

  • Apply Geolife datasets to Deepmove algorithm:

    python3 main_tf.py --use_geolife_data=True

    First, we need to get the dataset. You can find the detail from last item in Usage of MobilityPredictabilityUpperBounds. Then copy .npz file to /data dir. Afterwards, specify the filename in line51 of main_tf.py. Finally, run this command to get cross dataset accuracy.

MobilityPredictabilityUpperBounds

Requirements

  • python 3.8.5
  • numpy 1.20.2
  • scipy 1.6.3
  • healpy 1.14.0 # can't be installed on Windows OS
  • apsw 3.9.2.post1 # other versions can also work

Structure

already in MobilityPredictabilityUpperBounds dir

  • /ResultsLoP_replication # the dir which saves Heatmap.csv and Heatmap.pdf
  • GeolifeEntropyCalc.py # calucate entropy and build Heatmap.csv
  • GeolifeSymbolisation.py # convert Geolife data to symbolized positions
  • Graphing.py # plot Heatmap
  • Utils.py
  • ...

Usage

  • Pre-build the database cache files by running python3 GeolifeSymbolisation.py. This will take a significant amount of time as it builds the trajectory database from downloaded Geolife data set and pre-computes all required trajectory quantisations. If this is not done then it will be done on first run of the main code.
  • Run python3 GeolifeEntropyCalc.py. This computes the upper bounds for all spatiotemporal quantisations investigated in the paper and saves them to a CSV.
  • Run python3 Graphing.py. This plots the results as heatmaps.
  • Run python3 GeolifeSymbolisation.py --get_geolife_data=True --s=xx --t=xx. s denotes SpatialRes, and t denotes TemporalRes. Say we want to get part of Geolife dataset with SpatialRes=1000, TemporalRes = 1:00:00, we should run GeolifeSymbolisation.py --get_geolife_data=True --s=1000 --t=1:00:00. Then we can get a corresponding .npz file as "1000|1:00:00.npz". With this file, we could apply geolife datasets to DeepMove algorithm.
  • Run python3 GeolifeEntropyCalc.py --use_deepmove_dataset=True. This command will calculate upperbound probability for DeepMove dataset ranther than Geolife dataset.