Skip to content

njustkmg/TKDE21_HMM

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

This is the implementation of the paper Corporate Relative Valuation using Heterogeneous Multi-Modal Graph Neural Network.

Basic usage

python3 main.py --data_root ./sample_data --device cuda:0

Notice: The sample_data is a very small dataset for debug to make sure the training, validation and testing is runnable. This dataset is not a subset of our corporate relative valuation dataset, and the performance does not matter. To evaluate HMM model on other dataset, please refer to the following Custom Dataset section.

For details of all the arguments and help information,

python3 main.py -h

Requirements: pytorch scikit-learn

HMM.py: the HMM model.

cv_dataset.py: preprocess and build graph for the Corporate Relative Valuation dataset.

metrics.py: help functions for several metrics.

train_hmm.py: pytorch training code for HMM.

utils.py: help functions for calculating margin loss.

main.py: help and command line interface, detailed description of parameters.

Custom dataset

The data set used in this project will be announced after the article is published with the permission of the data owner.

We provide a detailed format here in case of usage on other datasets.

c_c.txt

(invest company, invested company, invest ratio)

('c1', 'c296', 1.0)
('c1', 'c302', 0.51)
('c1', 'c306', 1.0)
('c1', 'c300', 0.224)
('c1', 'c303', 1.0)

c_p.txt

(company, people, position id of this people in this company)

('c1', 'p1', [1.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0])
('c1', 'p4', [0.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 0.0])
('c1', 'p2', [0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0])

p_info.txt

people_id, people attribute

p7,0.221102,-0.8070350000000001,0.991509,0.807416,-0.005166,-0.487572,0.501344,-1.416632,0.628414,0.567874,0.360421,0.212601,0.36958,-0.47801499999999997,0.368847,-0.8551709999999999,-0.24893099999999999,0.567555,0.35022,-0.10656500000000001,-0.11441400000000002,-0.23879499999999998,-0.649752,0.722106,0.6276189999999999,-0.632771,-0.675759,-0.091075,1.2500879999999999,-0.808951,0.09175499999999999,0.000155,-0.646505,0.282987,-0.407227,-0.213216,-0.17617,0.181603,-1.038985,0.7805489999999999,0.092196,1.197251,-0.173746,-0.261974,-1.022464,-0.353187,0.538159,0.036936000000000004,-0.021117,0.826942
p8,0.32873600000000003,-1.1261709999999998,0.818144,0.40993100000000005,0.20371199999999998,-0.571954,0.441104,-1.072633,0.814242,0.429817,0.416253,0.29029499999999997,0.187653,-0.63593,0.547273,-0.9295540000000001,-0.28747,0.5485939999999999,0.383912,-0.158203,-0.085238,-0.080536,-0.75967,0.518534,0.532878,-0.452219,-0.629105,0.254504,1.04034,-0.967375,0.048351,0.16630799999999998,-0.8869879999999999,0.328864,-0.070145,0.061027,-0.077671,0.421911,-1.0601209999999999,0.887494,-0.0015630000000000002,1.148855,-0.492716,-0.569242,-0.47010799999999997,-0.383949,0.765278,0.281625,0.11590999999999999,1.1091010000000001

c_info.txt

company_id, company attribute

c8,0.03676470588235294,0.010101010101010102,0.5,0.07607607607607608,0.07407407407407407,0.0,0.10101010101010101,0.10101010101010101,0.14444444444444443,0.0,0.001001001001001001,0.05917159763313609,0.0,0.020202020202020204,0.020202020202020204,0.0,0.0,0.0,0.0,0.020202020202020204,0.020202020202020204,0.0,0.0,0.04040404040404041,0.04040404040404041,0.0,0.0,0.0,0.023023023023023025,0.020202020202020204,0.020202020202020204,0.0,0.0,0.5345345345345346,1.0,0.0,0.11983471074380166,0.29365079365079366,1.0,0.208,0.0,0.0,0.0,0.0,0.029411764705882353,0.8,0.01818181818181818,0.0,0.0,0.0,0.25,0.0,0.0,0.0,0.0,0.0,0.04040404040404041,0.04040404040404041,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,1.0,1.0,0.3333333333333333,0.2631578947368421,0.08,0.7,0.08708708708708708,0.0,0.005005005005005005,0.002002002002002002,0.011011011011011011,0.041928721174004195,0.657275,-0.862284,0.081813,-0.180645,0.016222,0.550884,0.255101,-0.552221,1.262831,0.915518,-0.199347,0.252942,-0.9568450000000001,0.480204,0.33599,-1.0000959999999999,-0.626958,0.263023,-0.655669,0.240542,-0.325152,-0.767717,-0.466624,0.8902389999999999,-0.37245300000000003,0.682639,0.001047,0.178925,-0.355698,0.738164,0.0033130000000000004,-1.058926,0.448704,0.812701,0.140457,0.07117899999999999,0.339116,0.724097,-0.639592,0.527254,0.31490999999999997,0.392828,0.40092300000000003,0.370839,0.137568,0.396802,1.189057,-0.051448,-0.005986,0.226675
c9,0.13970588235294118,0.08080808080808081,0.25,0.1001001001001001,0.8148148148148148,0.2702702702702703,0.030303030303030304,0.030303030303030304,0.37777777777777777,0.0,0.16616616616616617,0.1893491124260355,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.09090909090909091,0.09090909090909091,0.2222222222222222,0.035035035035035036,0.44344344344344344,1.0,0.0,0.004132231404958678,0.023809523809523808,1.0,0.994,0.0,0.0,0.0,0.0,0.058823529411764705,0.2,0.05454545454545454,0.0,0.0,0.16666666666666666,0.0,0.16666666666666666,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0018975332068311196,0.0,0.0,0.0,0.0,0.0,1.0,0.6111111111111112,0.0,0.10526315789473684,0.04,1.0,0.4694694694694695,0.07607607607607608,0.009009009009009009,0.005005005005005005,0.12412412412412413,0.03773584905660377,-0.682394,-0.7438739999999999,-0.080481,0.27123200000000003,0.41206800000000005,-0.49625600000000003,-0.087916,0.121876,1.084676,-0.09635099999999999,0.532459,0.32848299999999997,0.123735,-0.382792,0.292707,-0.973769,-0.510128,1.159096,-0.565704,0.186474,-0.09529800000000001,0.10375799999999999,-0.301349,0.5155420000000001,0.41341400000000006,0.12383599999999999,-0.45242899999999997,-0.5446,-0.8026260000000001,0.137468,-0.5049739999999999,-0.053739999999999996,-0.022221,0.315576,-0.150534,-0.510097,0.187574,0.52137,-0.398167,0.26422199999999996,-0.470272,0.630664,0.61446,1.325348,-1.107867,-0.347516,0.49695600000000006,-0.550531,-0.967125,-0.397032

c_label.txt

company_id, one-hot encoding of value, one-hot encoding of area

c11,1.0,0.0,0.0,0.0,1.0,0.0,0.0,0.0,0.0,0.0,0.0
c12,1.0,0.0,0.0,0.0,0.0,0.0,0.0,1.0,0.0,0.0,0.0
c13,1.0,0.0,0.0,0.0,1.0,0.0,0.0,0.0,0.0,0.0,0.0

The first line of this file is ignored.

splits/

This folder contains several splits of dataset with different training ratio.

split_10.pkl
split_30.pkl
split_50.pkl
split_70.pkl

Each split is a pickle file contains a dictionary {"train_idx": train_idx, "test_idx": test_idx}

The splits can be modified easily in train_hmm.py

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages