"data_analysis_interval"] = data_analysis_interval
anay_data: AnalyticalData = AnalyticalData()
anay_data.settings_info_dict = game_raw_data.settings_info_dict
anay_data.name_datalist_dict["mean"] = [0]
anay_data.name_datalist_dict["standard_deviation"] = [0]
if is_save_chart_data or is_save_analytica_data or is_save_raw_data:
CommonUtil.make_dir(save_dir_path)
CommonUtil.save_json_to_file(game_raw_data.__dict__,
save_dir_path + "/settings_info.txt")
env = gym.make('CartPole-v0')
tabular_agent: BinaryOutputTabularAgent = BinaryOutputTabularAgent(
initial_epsilon, epsilon_decay, episodes_per_drop, learning_rate,
discount_rate)
for episode_idx in range(episodes_to_run):
if is_print_episode_idx:
CommonUtil.print_on_same_line("episode: " + str(episode_idx + 1) +
".\t")
cartpole_state = env.reset()
is_game_over = False
total_reward = 0
time_step = 0
while not is_game_over:
time_step += 1
tmp_interval_mean_list = [0]
tmp_interval_stddev_list = [0]
tmp_evaluation_msg_list = []
if is_save_chart_data or is_save_analytica_data:
CommonUtil.make_dir(save_dir_path)
CommonUtil.save_json_to_file(game_raw_data.__dict__,
save_dir_path + "/settings_info.txt")
print("start running episodes")
for episode_idx in range(num_of_iterations):
if is_print_episode_idx:
CommonUtil.print_on_same_line("episode " + str(episode_idx + 1) +
". ")
exploration_rate = BinaryOutputTabularAgent.derive_explore_rate(
episode_idx, initial_epsilon, epsilon_decay, episodes_per_drop)
game_data.general_data_list.append(exploration_rate)
# data analysis section
if episode_idx == 0 or (episode_idx + 1) % update_interval == 0:
print(f"Explore rate: {exploration_rate}")
start_interval: int = episode_idx - update_interval + 2
start_interval_str: str = CommonUtil.format_int_to_str_length(
start_interval, len(str(num_of_iterations)))
end_interval_str: str = CommonUtil.format_int_to_str_length(
episode_idx + 1, len(str(num_of_iterations)))
plt = PlotUtil.plot(plot_id=1,
plot_title="Exploration Rate Change",
x_label="Episodes",
y_label="Exploration Rate",