This code aims to optimize waterfall auction strategy.
The code is based on Viterbi path (Viterbi, 1967; Rabiner, 1989)
Required packages: os, math, logging, pickle, pprint, copy, numpy, pandas, scipy
The repository is organized as follows:
classes - all the necessary classes
data - the synthetic datasets and their generator
models - the Viterbi algorithm
output - the relevant results
findAllWaterfalls.py generates all possible watterfall for MatrixM.csv with 1<=r<=9 (waterfall length)
*Quick start*: run main.py