text="""<p><b>{}</b> rivers with stations at risk, the top 5 is tabulated below.</p>""".format(
len(rivers_with_station(risky_stations))),
width=600
)
warning_text2.sizing_mode = 'scale_width'
risky_rivers = rivers_by_station_number(risky_stations, 5)
risky_rivers_source = ColumnDataSource(
data=dict(name=[i[0] for i in risky_rivers], num=[i[1] for i in risky_rivers])
)
risky_river_table_columns = [
TableColumn(field="name", title="River Name"),
TableColumn(field="num", title="Number of Risky Stations"),
]
risky_river_table = DataTable(source=risky_rivers_source, columns=risky_river_table_columns, width=500, height=140)
risky_river_table.sizing_mode = 'scale_width'
# Risky towns
risky_towns = []
key_station_in_cluster = []
mean_levels = []
for label, s in label_to_stations.items():
cluster_levels = np.array([i.relative_water_level() for i in s])
mean_levels.append(cluster_levels.mean())
key_station_in_cluster.append(s[np.argmax(cluster_levels)])
for i in s:
risky_towns.append(i.town)
risky_indx = risky_name_to_indx[i.name]
risky_source.data['risk'][risky_indx], risky_source.data['alpha'][risky_indx] = 'High', 1.0
risky_towns = set(risky_towns) # to find the total number of risky towns
def fill_experiment_table():
"""Populates the table that displays the results from previously ran
experiments. The results are stored as yaml files in the directory
that the global variable EXPERIMENT_SERIALIZATION_DIR points to.
"""
global data_table, table_source, table_data_df
os.chdir(EXPERIMENT_SERIALIZATION_DIR)
experiment_files = glob.glob("*.yml")
names = [''] * len(experiment_files)
Ls = [0] * len(experiment_files)
reward_types = [''] * len(experiment_files)
window_size = [''] * len(experiment_files)
window_step = [''] * len(experiment_files)
seq_or_con = [''] * len(experiment_files)
for i, filename in enumerate(experiment_files):
with open(filename, 'r') as ymlfile:
experiment_data = yaml.safe_load(ymlfile)
names[i] = filename
Ls[i] = experiment_data['L']
reward_path = experiment_data['reward_path']
if 'threshold' in reward_path:
reward_types[i] = 'threshold' + reward_path.split('threshold_')[
1][0:3]
elif 'top' in experiment_data['reward_path']:
reward_types[i] = 'top'
else:
if 'pear' in reward_path:
reward_types[i] = 'continous-pearson'
elif 'MI_n1' in reward_path:
reward_types[i] = 'continous-MI-normalized'
else:
reward_types[i] = 'continous-MI'
window_size[i], window_step[i] = re.findall(
r'\d+', reward_path)[0:2]
seq_or_con[i] = reward_path.split('/')[-1][0:3]
table_data = dict(
names=names,
seq_or_con=seq_or_con,
Ls=Ls,
reward_types=reward_types,
window_size=window_size,
window_step=window_step,
)
table_data_df = pd.DataFrame(data=table_data)
table_data_df = table_data_df.sort_values(
['seq_or_con', 'Ls', 'window_size', 'window_step', 'reward_types'])
table_source = ColumnDataSource(data=table_data_df)
table_source.selected.on_change('indices', load_experiment)
columns = [
TableColumn(field="names", title="Experiment"),
TableColumn(field="seq_or_con", title="Sequential/ concurrent"),
TableColumn(field="Ls", title="L"),
TableColumn(field="reward_types", title="Reward Type"),
TableColumn(field='window_size', title="Window Size"),
TableColumn(field='window_step', title="Window Step"),
]
data_table = DataTable(source=table_source, columns=columns, height=200)
data_table.sizing_mode = 'stretch_width'
global pol_table, pol_source
columns_pol_table = [
TableColumn(field="pol_name", title="Name"),
TableColumn(field="overall_regret", title="Overall regret")
]
pol_source = ColumnDataSource(dict(pol_name=[], overall_regret=[]))
pol_source.selected.on_change('indices', plot_policy)
pol_table = DataTable(
source=pol_source, columns=columns_pol_table, height=300)
pol_table.sizing_mode = 'stretch_width'
