- setup github actions
- add fail safe for function input parameters
- added function to find a file path based on code and replace parameters
- updated web links
- reduced execution time by reducing certain examples
- used R version 4.4.1
- removed as.vector from example code to fix error in R-devel builds
- increased min R version to 3.6.0
- add second parameter to simulation function
aml_simulation - option to use full columns for model training when selecting 0 as a parameter
num_cols_used - add suppress messages option during
readr::read_csv()function calls - fail safe in
aml_collect_datafunction will delete already recorded rds file if it has different amount of columns - add new function
util_find_pidto find the PIDs of the terminal.exe application - function
mt_stat_transfis now using a rule to assign 3 market type classes - rewrite function
mt_make_modelwith the same philosophy as inaml_make_model
- designed to test different inputs
- fix bug in not allowing use of full input to the model
- create new option with fix neural network structure
- add parameter to split data with specific n of rows
- add parameter num_bars_ahead, default is 34 bars
- add options for deep learning function parameters search,
num_epochs,fixed_nn_struct - add option
objective_testto allow strategy test inside the function to select best model based on specific objective
- more accurate calculation of Label column using tick data
- added fail safe scenarios
- add option in function
aml_consolidateresults.Rto help understand overall model performances and log results - removed unused functions and datasets
- add lifecycle badges
- added fail safe stops in function
check_if_optimize.R - experimenting with time series statistical transformation adding functions
mt_stat_transfandmt_stat_evaluate. Learning opportunity around clustering for automatic market type generation - added fail safe and change the way how to find control parameters for Reinforcement Learning to avoid slow code execution
- solving warnings by changing example in mt_make_model function
- change function mt_make_model to read manually checked dataset
- update documentation
- solving warnings
- added parameter min_perf to adjust model performance
- consider tick value in the calculations of strategy testing
- added option to balance classes in mt_make_model function
- add and delete deprecated functions
- change names of functions to reflect the purpose: mt - market type, rl- reinforcement learning, util - utility
- change example methods to unify them across functions
- align names of the variables across several functions
- remove warnings caused by new dplyr functionalities .groups
- remove tidyr from imports to solve Note
- Added extended package features in the Readme file
- Re-write functions aml*.R
- aml_collect.. will keep the time index
- aml_make_model ... will build a model
- aml_score... is designed to use model and predict
- aml_test... is designed to perform a strategy test
- mt_make_model .. rewrite in the similar fashion as in aml function
- examples and tests are available
- Add badges 'Maturing', 'CRAN status'
- Add Methods source in Description file
- Add packages name syntax e.g.
'readr::write_csv' - Solve Namespace dependencies notes
- check of examples marked as /donttest, rerun with --run-donttest
- removed some examples with /donttest
- added functions
encrypt_api_key.R,mt_make_model.R - minor enhancements
- simplification of R/h2o updates, function
aml_make_model()may conditionally force model update - added function to simplify creation of configuration files to launch MT4 platform
- added utility
util_generate_password()to generate passwords e.g. for MT4 platform
- new function
opt_aggregate_results()to aggregate results of the trading robots performance - new function
opt_create_graphs()to create graphs of the trading robots performance - added more tests and removal of warning by using
ungroup()function
- removal of duplicate rows in the
aml_collect_data()function - added a test for
to_m()function - add warning to indicate if 'zeroes' are present in the data file column(s)
- removing dependency from tidyverse
- added more datasets (files) to test functionality
- building a functionality to create separate models for each asset (adding functions with prefix aml_xxx_xxx.R)
- build more simple functions to collect data, build model, score data to predict
- added tests for several functions
- function aml_make_model() will be finding the best neural network structure automatically using 6 random structures
- fixing of warnings due to column name assignments
- corrected bug on file naming for Terminal 3 on function record_policy_mt.R
- better defined R dependency
- corrected file writing example to be aligned with CRAN Policy
- added test coverage integration
- corrected file writing example to be aligned with CRAN Policy
- corrected according to the CRAN team comments
- Initial release
- Passing all checks