SCPP: Shared Cascade Poisson Process

Unofficial Python implemetation of:
Discovering latent influence in online social activities via shared cascade poisson processes, Iwata, T., Shah, A. and Ghahramani, Z. In KDD. 2013.

DEMO

• python3 main.py --retail
  to process: https://www.kaggle.com/regivm/retailtransactiondata

Command line options

• --input_fn: csv file name
• --output_dir: name of an output directory (default: 'out/tmp')
• --gamma: initial value of the decay parameter (default: 1)
• --beta: initial value of the Dirichlet paramter (default: 1)
• --a: hyperparameter for Gamma distribution (default: 1)
• --b: hyperparameter for Gamma distribution (default: 1)
• --tol: tolerance for early stopping (default: 100)
• --max_iter: maximum number of the EM iteration (default: 20)
• --time_col: column name of timestamps (default: 'date')
• --item_col: column name of items (default: 'item')
• --user_col: column name of users (default: 'user')
• --n_sample: number of samples/events from original records (default: 100000)
• --sampling_rate: sampling rate when encoding timestamps (default: 'D')

Reference

• Minka, T., 2000. Estimating a Dirichlet distribution.