taxi-demand-modeling

1. Reference

• Meng-Fen Chiang, Tuan-Anh Hoang, Ee-Peng Lim:, "Where are the passengers?: a grid-based gaussian mixture model for taxi bookings." SIGSPATIAL/GIS 2015

2. Input Format

• Each booking is put in a file. Each line in the file is of the format:
  ...
  <date_i>,<time-of-a-day_i>,<latitude_i>,<longitude_i>,<lat-grid-id_i>,<lon-grid-id_i>
  ...
  e.g., 2014-05-06,503.0,1.321490,103.908600,7,34
• File path: rootdir/grabTaxi/input/sg_bookings_jul_sep.txt_wday_b100.csv

3. Output Format

• Loglikelihood: rootdir/grabTaxi/output/logLik(k1)_(k2)_wday_jul_sep_d(time-slot).csv
  where (k1), (k2) and (time-slot) are smallest num K, largest num K and time window of interest, which are specified during running the model
• Parameters: rootdir/grabTaxi/output/GGMM/ggmm_(day)_jul_sep_k_w(t1)_w(t2) where (t1)/(t2) is the starting/ending time windows, which is specified during running the model.
  • mixture weights of K Gaussians of each cell
  • mu of K Gaussians of each cell
  • covariance matrix of K Gaussians of each cell
    e.g.,
    ...
    pi_0=[ 0.294605809129 0.634854771784 0.0705394190871 ]
    clusters = 3
    mu_0=[ 1.30096634019 103.828820661 360.28973179 ]
    sigma_0=[ [4.060461561835867e-05, 0.0, 0.0] [0.0, 0.0046409619060315144, 0.0] [0.0, 0.0, 438.4028687300076] ]
    ...