Code

Machine learning#1

Code

Open

alirezamoosio wants to merge 113 commits intoepfldata:masterepfldata/SimulationEngine:masterfrom alirezamoosio:machine-learningalirezamoosio/economic_simulations:machine-learningCopy head branch name to clipboard

+274,104-3,220

Commits

Commits on Jul 24, 2019

• Resolved compile errors and made the project runnable

  

  mousavi

  committed
  71d5db7

  Copy full SHA for 71d5db7

• connected bayesian optimizer of python to bo.Main

  

  mousavi

  committed
  bdd8119

  Copy full SHA for bdd8119

Commits on Jul 25, 2019

• added a machinery for input/output generation and evaluation

  

  mousavi

  committed
  e97a47a

  Copy full SHA for e97a47a

• change scala call from python

  vrupp

  committed
  0c1f154

  Copy full SHA for 0c1f154

• resolved merge conflicts sbt call

  vrupp

  committed
  3084af9

  Copy full SHA for 3084af9

• added scaladoc comments for BOUtil

  

  mousavi

  committed
  a15f26d

  Copy full SHA for a15f26d

• call sbt package

  vrupp

  committed
  ff25c6e

  Copy full SHA for ff25c6e

• call sbt package

  vrupp

  committed
  145a6c5

  Copy full SHA for 145a6c5

Commits on Jul 26, 2019

• added params to BO

  vrupp

  committed
  ab35e31

  Copy full SHA for ab35e31

Commits on Jul 29, 2019

• added the basic tool for calculating adapted lyapunov exponent

  

  mousavi

  committed
  6bdd2d3

  Copy full SHA for 6bdd2d3

• refactored params into simulation class

  vrupp

  committed
  a38758f

  Copy full SHA for a38758f

• added simple visualization tools

  

  mousavi

  committed
  cfdf226

  Copy full SHA for cfdf226

Commits on Jul 30, 2019

• read params from json

  vrupp

  committed
  a94b2ef

  Copy full SHA for a94b2ef

• added json support to bayesian_optimizer.py

  

  mousavi

  committed
  40473e9

  Copy full SHA for 40473e9

Commits on Jul 31, 2019

• added a train/test model for evaluating bayesian optimization performance

  

  mousavi

  committed
  85d49f6

  Copy full SHA for 85d49f6

• added a train/test model for evaluating bayesian optimization performance

  

  mousavi

  committed
  97c4199

  Copy full SHA for 97c4199

• Merge remote-tracking branch 'origin/master'

  

  mousavi

  committed
  de18b53

  Copy full SHA for de18b53

• added params to simulation

  vrupp

  committed
  d04063d

  Copy full SHA for d04063d

• changed error function

  vrupp

  committed
  b097d13

  Copy full SHA for b097d13

Commits on Aug 2, 2019

• added factory params

  vrupp

  committed
  98ee926

  Copy full SHA for 98ee926

Commits on Aug 5, 2019

• tuned the params for meaningful output, added two-mode plotting from sbt command line

  

  mousavi

  committed
  e2d2cc1

  Copy full SHA for e2d2cc1

Commits on Aug 6, 2019

• fixed error calculation and log with buffer

  vrupp

  committed
  34389a6

  Copy full SHA for 34389a6

• added partial-parameter optimization to python

  

  mousavi

  committed
  2c645f1

  Copy full SHA for 2c645f1

• plot multiple lines with same intervall, fix param intervalls for BO

  vrupp

  committed
  d0a8cc6

  Copy full SHA for d0a8cc6

Commits on Aug 7, 2019

• simplified optimization process and removed inputs

  

  mousavi

  committed
  0987cfa

  Copy full SHA for 0987cfa

• modified bayesian optimization loop

  

  mousavi

  committed
  925ebbf

  Copy full SHA for 925ebbf

Commits on Aug 8, 2019

• added params for gender and capital, normalized viz output

  vrupp

  committed
  cb1ec06

  Copy full SHA for cb1ec06

• removed output ranges

  vrupp

  committed
  31423fd

  Copy full SHA for 31423fd

Commits on Aug 9, 2019

• added observables, fixed bug empty pl

  vrupp

  committed
  90c74d9

  Copy full SHA for 90c74d9

• fixed viz bugs

  vrupp

  committed
  1becf3d

  Copy full SHA for 1becf3d

• write data to csv before plotting

  vrupp

  committed
  e86e3f2

  Copy full SHA for e86e3f2

• set higher resolution for viz

  vrupp

  committed
  e7f2367

  Copy full SHA for e7f2367

Commits on Aug 12, 2019

• generate csv for train set

  vrupp

  committed
  58ed859

  Copy full SHA for 58ed859

• added the first implementation of rnn

  

  mousavi

  committed
  0c88406

  Copy full SHA for 0c88406

• created time series train csv data

  vrupp

  committed
  23bc72c

  Copy full SHA for 23bc72c

• added training from csv to rnn.py

  

  mousavi

  committed
  aec76f5

  Copy full SHA for aec76f5

• created csv for x and y variable

  vrupp

  committed
  a864170

  Copy full SHA for a864170

Commits on Aug 13, 2019

• added cross fold validation

  vrupp

  committed
  91195b0

  Copy full SHA for 91195b0

• made it able to run on gpu

  vrupp

  committed
  a8bf578

  Copy full SHA for a8bf578

Commits on Aug 14, 2019

• added simple version of rnn for test/debug

  

  mousavi

  committed
  8d2d9fc

  Copy full SHA for 8d2d9fc

Commits on Aug 23, 2019

• removed all unused classes and methods

  vrupp

  committed
  a0af531

  Copy full SHA for a0af531

• fixed salary bug

  vrupp

  committed
  4cbb3c9

  Copy full SHA for 4cbb3c9

• added factories and params for their pls

  vrupp

  committed
  2817c0c

  Copy full SHA for 2817c0c

Commits on Aug 26, 2019

• added agent.py

  

  mousavi

  committed
  b00dae5

  Copy full SHA for b00dae5

• create csv for training a person agent

  vrupp

  committed
  abed14c

  Copy full SHA for abed14c

• added env.py

  

  mousavi

  committed
  25e1869

  Copy full SHA for 25e1869

• added example for using enivornment/agents

  

  mousavi

  committed
  d43c827

  Copy full SHA for d43c827

• scaled input

  vrupp

  committed
  e3b5036

  Copy full SHA for e3b5036

Commits on Aug 27, 2019

• implemented a high-level(relationship graph)/low-level(bunch of networks) conversion

  

  mousavi

  committed
  1b9cd5d

  Copy full SHA for 1b9cd5d

Commits on Aug 28, 2019

• mapped values between env and graph

  vrupp

  committed
  7fb4699

  Copy full SHA for 7fb4699

Commits on Aug 29, 2019

• correlation matrix for agent

  vrupp

  committed
  065ac9b

  Copy full SHA for 065ac9b

• derivative matrix

  vrupp

  committed
  1a366b1

  Copy full SHA for 1a366b1

Commits on Aug 30, 2019

• Setting up the experiment: cleaning up parameters

  

  mousavi

  committed
  70cdecd

  Copy full SHA for 70cdecd

• Setting up the experiment: speicalized factories into different agent types

  

  mousavi

  committed
  72da1f6

  Copy full SHA for 72da1f6

Commits on Aug 31, 2019

• Setting up the experiment: added global statistics for joint training

  

  mousavi

  committed
  d6a78fc

  Copy full SHA for d6a78fc

Commits on Sep 1, 2019

• implemented group train

  vrupp

  committed
  7f6224a

  Copy full SHA for 7f6224a

Commits on Sep 2, 2019

• added two testing scenarios for the network model

  

  mousavi

  committed
  c5eda0c

  Copy full SHA for c5eda0c

• changed double type of variables to function

  

  mousavi

  authored and

  vrupp

  committed
  8b40281

  Copy full SHA for 8b40281

• updated generation and evaluation data for BO

  vrupp

  committed
  6684340

  Copy full SHA for 6684340

Commits on Sep 3, 2019

• renamed params and added params in BO python file

  vrupp

  committed
  2f63ce9

  Copy full SHA for 2f63ce9

• added single_nn.py for comparison with our network

  

  mousavi

  committed
  8180a38

  Copy full SHA for 8180a38

• fixed typo

  vrupp

  committed
  b6bcd0e

  Copy full SHA for b6bcd0e

• added the option of configuring low-level network while defining an agent

  

  mousavi

  committed
  2bb77b7

  Copy full SHA for 2bb77b7

• Merge remote-tracking branch 'origin/master'

  

  mousavi

  committed
  068efd6

  Copy full SHA for 068efd6

Commits on Sep 4, 2019

• set up environment configuration for our simulation plus solo training/testing

  

  mousavi

  committed
  3dcbccc

  Copy full SHA for 3dcbccc

Commits on Sep 5, 2019

• added analysis tool for simdata/debugged random initialization bug

  

  mousavi

  committed
  e8b5cf7

  Copy full SHA for e8b5cf7

• learn input values with tf optimizer

  vrupp

  committed
  0d25b97

  Copy full SHA for 0d25b97

• devided state indicators to (initializable) variables and (over time) constants

  

  mousavi

  committed
  b2696ea

  Copy full SHA for b2696ea

• added saving and loading ability to nn models

  

  mousavi

  committed
  2ec3975

  Copy full SHA for 2ec3975

• implemented group train for simulation

  vrupp

  committed
  8f3656a

  Copy full SHA for 8f3656a

Commits on Sep 6, 2019

• group train for simulation

  vrupp

  committed
  f7a74cd

  Copy full SHA for f7a74cd

• implemented input learning for simulation

  vrupp

  committed
  81fce6f

  Copy full SHA for 81fce6f

Commits on Sep 9, 2019

• implemented prediction over time/added scale save-load

  

  mousavi

  committed
  b3e41ab

  Copy full SHA for b3e41ab

• resolved a bug with global statistics (Source is a sim but not an agent)

  

  mousavi

  committed
  a703278

  Copy full SHA for a703278

• Partially updated Readme

  alirezamoosio

  authored
  e293602

  Copy full SHA for e293602

• python code done! except debugging

  

  mousavi

  committed
  6ce7e52

  Copy full SHA for 6ce7e52

• added recursion depth for better debugging

  vrupp

  committed
  a1a1b5d

  Copy full SHA for a1a1b5d

Commits on Sep 10, 2019

• added python docstring for some files (not complete)

  

  mousavi

  committed
  bbeca8c

  Copy full SHA for bbeca8c

• Update README.md

  alirezamoosio

  authored
  60c60f5

  Copy full SHA for 60c60f5

• Update README.md

  alirezamoosio

  authored
  47259ff

  Copy full SHA for 47259ff

• Update README.md - fixing link bugs

  

  alirezamoosio

  authored and

  mousavi

  committed
  e82ab97

  Copy full SHA for e82ab97

• Merge remote-tracking branch 'origin/master'

  Show description for 3bff5e7
  

  mousavi

  committed
  3bff5e7

  Copy full SHA for 3bff5e7

• fixed bug with copying agents' constants

  vrupp

  committed
  827cd3e

  Copy full SHA for 827cd3e

• changed BO for new json format and equalized param ranges

  vrupp

  committed
  be73f6b

  Copy full SHA for be73f6b

Commits on Sep 11, 2019

• simplified bayesian optimization

  vrupp

  committed
  efb0364

  Copy full SHA for efb0364

• fixing typo

  vrupp

  committed
  2c1b7d7

  Copy full SHA for 2c1b7d7

• saving intermediate results BO

  vrupp

  committed
  46a9fd6

  Copy full SHA for 46a9fd6

• Another update to the README.md!

  alirezamoosio

  authored
  3b1ea8b

  Copy full SHA for 3b1ea8b

• added learning rate for group training and input learning

  

  mousavi

  committed
  567974d

  Copy full SHA for 567974d

• added supplementary files and changed readme

  

  mousavi

  committed
  0093912

  Copy full SHA for 0093912

• write bo target in result json

  vrupp

  committed
  70049c7

  Copy full SHA for 70049c7

• refactored network and bo for supplementary directory

  vrupp

  committed
  04541a0

  Copy full SHA for 04541a0

• moved params.json to supplementary

  vrupp

  committed
  d7621d6

  Copy full SHA for d7621d6

Commits on Sep 12, 2019

• added sigma variables to data generation and network configuration

  

  mousavi

  committed
  0b0ae97

  Copy full SHA for 0b0ae97

• tuned single_nn's config and added 0 sd support

  

  mousavi

  committed
  5aa12a8

  Copy full SHA for 5aa12a8

• wrote results of input learning in json file

  vrupp

  committed
  3bd9d62

  Copy full SHA for 3bd9d62

• minimum 2 agents per type

  vrupp

  committed
  aacc814

  Copy full SHA for aacc814

• added batch-prediction

  

  mousavi

  committed
  0e7e7ea

  Copy full SHA for 0e7e7ea

• removed todos and added docstring

  vrupp

  committed
  7e18cdf

  Copy full SHA for 7e18cdf

• fixed the 0 sd problem of single_nn.py

  

  mousavi

  committed
  c940bc5

  Copy full SHA for c940bc5

• Merge branch 'master' of https://github.com/alirezamoosio/economic_simulations

  

  mousavi

  committed
  2cace38

  Copy full SHA for 2cace38

Commits on Sep 13, 2019

• adding datasets with 3000 lines

  vrupp

  committed
  db8762d

  Copy full SHA for db8762d

• models with new dataset

  vrupp

  committed
  d334845

  Copy full SHA for d334845

Commits on Sep 16, 2019

• comment assertion because of imprecisely rounding

  vrupp

  committed
  4797c1e

  Copy full SHA for 4797c1e

Commits on Sep 19, 2019

• added global stat scaling for error measurement in scala

  

  mousavi

  committed
  b8fc91a

  Copy full SHA for b8fc91a

Commits on Sep 27, 2019

• splitted dataset into differently generated train and test data

  

  mousavi

  committed
  bd5f5c1

  Copy full SHA for bd5f5c1

• splitted dataset into differently generated train and test data

  

  mousavi

  committed
  87acd82

  Copy full SHA for 87acd82

• Merge remote-tracking branch 'origin/master'

  

  mousavi

  committed
  99f33c4

  Copy full SHA for 99f33c4

Commits on Sep 28, 2019

• changed all group metrics to mae for consistency

  

  mousavi

  committed
  b48de67

  Copy full SHA for b48de67

Commits on Sep 29, 2019

• fixed an input-learning bug, the same input parameters for different agents were treated as different parameters!

  

  mousavi

  committed
  9c95be4

  Copy full SHA for 9c95be4

Commits on Sep 30, 2019

• added the ability to run the simulation directly from bo.Main

  

  mousavi

  committed
  62abfb3

  Copy full SHA for 62abfb3

Commits on Oct 3, 2019

• moved files to ml subdirectory

  

  mousavi

  committed
  3bd7d43

  Copy full SHA for 3bd7d43

Commits on Nov 25, 2019

• moved everything to machine-learning branch

  

  mousavi

  committed
  bb2af50

  Copy full SHA for bb2af50