ClimbsRocks/machineJS

Issues

• MultinomialNB fails when features have negative values

  #176 opened 8 years ago by sunnmy
  2

• List index out of range

  #179 opened 8 years ago by ittailup
  0

• Windows multithreading doesn't work

  #178 opened 8 years ago by TalhaAsmal
  0

• Questions on hyperparameter distributions and validation percentage

  #174 opened 8 years ago by ClimbsRocks
  0

• Store validation score in file for ensemble learning

  #173 opened 8 years ago by kprimice
  2

• Something wrong with data formatter?

  #153 opened 8 years ago by TasmaniaKrama
  11

• stop training after a single round if a model is performing very poorly

  #172 opened 8 years ago by ClimbsRocks
  0

• verify that we properly stop training algorithms when they haven't been performing for three rounds

  #143 opened 8 years ago by ClimbsRocks
  2

• new installation script

  #161 opened 8 years ago by ClimbsRocks
  1

• investigate how we handle skipping an algo

  #169 opened 8 years ago by ClimbsRocks
  1

• for the validationRound, write the score from the hyperparameter search, not the final training run

  #170 opened 8 years ago by ClimbsRocks
  2

• Error in installation guide

  #171 opened 8 years ago by mmajewsk
  2

• bump down the boost for numRounds for linear models

  #168 opened 8 years ago by ClimbsRocks
  1

• shut everything down if there's an error with data-formatter

  #156 opened 8 years ago by ClimbsRocks
  1

• reduce number of times we train each model

  #159 opened 8 years ago by ClimbsRocks
  1

• tag some of these issues as quick wins and steadily knock them out

  #162 opened 8 years ago by ClimbsRocks
  0

• verify we're running RandomizedSearchCV only once for each model during the ensembler round

  #166 opened 8 years ago by ClimbsRocks
  1

• Issue after data-formatter

  #154 opened 8 years ago by kprimice
  2

• Classifier: output proba instead of category

  #155 opened 8 years ago by kprimice
  1

• send an email out with updates

  #167 opened 8 years ago by ClimbsRocks
  0

• update verbiage from classifiers to estimators now that we support regression

  #137 opened 8 years ago by ClimbsRocks
  0

• write all trained algo's parameters to file

  #134 opened 9 years ago by ClimbsRocks
  0

• update documentation to make it stupid easy to get started with

  #164 opened 8 years ago by ClimbsRocks
  0

• generally simplify documentation

  #165 opened 8 years ago by ClimbsRocks
  0

• add in a table showing an example dataDescription for the .csv files

  #163 opened 8 years ago by ClimbsRocks
  0

• set default to only use linear models for nlp

  #157 opened 8 years ago by ClimbsRocks
  1

• add in more linear models

  #158 opened 8 years ago by ClimbsRocks
  0

• Blog Post Series: How to customize machineJS

  #160 opened 8 years ago by ClimbsRocks
  0

• Cannot find pySetup/utils.py

  #149 opened 8 years ago by eanie
  20

• Persist everything to disk so a trained instance can be loaded

  #152 opened 8 years ago by ClimbsRocks
  1

• future ideas:

  #151 opened 8 years ago by ClimbsRocks
  0

• allow custom validation splits, like numer.ai uses

  #150 opened 8 years ago by ClimbsRocks
  1

• TImeseries data?

  #148 opened 8 years ago by JohnnySheffield
  2

• link in "explaining machineJS to my parents" series

  #147 opened 8 years ago by ClimbsRocks
  0

• for debugging purposes, load in the raw csv

  #146 opened 8 years ago by ClimbsRocks
  0

• improve accuracy

  #145 opened 8 years ago by ClimbsRocks
  1

• have intentional pacing of SIGINT signals

  #144 opened 8 years ago by ClimbsRocks
  0

• add support for multiple categories in the output column

  #140 opened 8 years ago by ClimbsRocks
  2

• give the user the option of returning a multi-category matrix, not just a single column

  #142 opened 8 years ago by ClimbsRocks
  2

• the MLP appears to blow up huge amounts of memory during multi-category classification

  #141 opened 8 years ago by ClimbsRocks
  0

• tweak search parameters

  #139 opened 8 years ago by ClimbsRocks
  0

• have amazing documentation

  #138 opened 8 years ago by ClimbsRocks
  0

• add in polynomial interactions between features

  #129 opened 9 years ago by ClimbsRocks
  1

• only run the MLP if the user has ^v.0.18.0 of sklearn installed

  #133 opened 9 years ago by ClimbsRocks
  1

• create installation script

  #135 opened 9 years ago by ClimbsRocks
  1

• packages to install in installation script

  #136 opened 9 years ago by ClimbsRocks
  1

• figure out which parameter is throwing all the warnings with MLP

  #132 opened 9 years ago by ClimbsRocks
  0

• once sklearn 0.18 is released, handle the deprecationWarnings

  #131 opened 9 years ago by ClimbsRocks
  0

• try adding warm_start as a param in makeClassifiers.py

  #128 opened 9 years ago by ClimbsRocks
  1

• move around logic for getting the classifierList until after data-formatter has run

  #130 opened 9 years ago by ClimbsRocks
  1