exarl.utils.analyze_reward

Module Contents

Functions

read_data(filename, rank)

The function reads csv-based learning data from the given log file into a pandas frame for use in plotting and result analysis.

save_reward_plot()

Creates and saves a Rolling Total Reward (y-axis) by Relative Time (x-axis) plot based on .log files written by EXARL for each rank.

Attributes

logger

exarl.utils.analyze_reward.logger
exarl.utils.analyze_reward.read_data(filename, rank)

The function reads csv-based learning data from the given log file into a pandas frame for use in plotting and result analysis.

Parameters
  • filename (string) – csv file of log data from EXARL

  • rank (integer) – MPI rank number

Returns

pandas.DataFrame – contains data extracted from the csv-based EXARL learning log file, except for current_state and next_state fields, and with the addition of a rank field.

exarl.utils.analyze_reward.save_reward_plot()

Creates and saves a Rolling Total Reward (y-axis) by Relative Time (x-axis) plot based on .log files written by EXARL for each rank. It saves the plot in the results directory named by the output_dir run parameter in a subdirectory /Plots/reward_plot.png. It then tries to print the plot to the terminal.