slider_active_toggle.label = "slider not active"
slider_active_toggle.button_type = "default"
# assign callbacks
play_button = Toggle(label="PLAY",
button_type="primary",
width=100,
callback=callback_play)
time_slider.on_change('value', callback_time_slider)
slider_active_toggle.on_click(callback_toggle_slider_activity)
pattern_select.on_change('active', callback_pattern_selection)
aggregate_select.on_change('active', callback_aggregation_selection)
groupby_select.on_change('active', callback_groupby_selection)
type_filter.on_change('value', callback_type_filter)
geo_source.on_change('selected', callback_map_selection)
select_all_types_button.on_click(callback_select_all_types)
## end callbacks
# headers
map_head = Div(text="<h2>Spatial distribution of incidents</h2>",
css_classes=["plot-head"],
width=LEFT_COLUMN_WIDTH)
ts_head = Div(text="<h2>Time distribution of incidents</h2>",
css_classes=["plot-head"],
width=RIGHT_COLUMN_WIDTH)
status_available_style = {"font-size": "8pt", "color": "green"}
status_unavailable_style = {"font-size": "8pt", "color": "red"}
status = Div(text="""<i>Status: at your service</i>""",
style={
"font-size": "8pt",
def output_res():
with open(r'./twgeo_easy_rates.json', 'r') as file:
json_str = file.read()
geo_source = GeoJSONDataSource(geojson=json_str)
twgeo_json = json.loads(json_str)
pred_4week = pd.read_csv('prediction_4week.csv')
areas = ['南區', '北區', '東區', '高屏區', '台北區', '中區']
models = ['CatBoost', 'XGB', 'ConvLSTM', 'GRU', 'LGBM', 'truth']
color_list = ['black', 'red', 'blue', 'green', 'cyan', 'purple']
area_dict = {}
for idx, area in enumerate(areas):
area_dict[area] = {}
temp = pd.read_csv('./data/{}.csv'.format(area))
area_dict[area]['df'] = temp
area_dict[area]['DT'] = pd.to_datetime(temp['DT'].values)
area_idx = []
for i in range(22):
if twgeo_json['features'][i]['properties']['area'] == area:
area_idx.append(i)
area_dict[area]['area_idx'] = area_idx
def fig_config(fig):
fig.grid.grid_line_color = None
fig.toolbar.logo = None
fig.toolbar_location = None
fig.legend.location = "top_left"
fig.legend.click_policy = "hide"
fig.xaxis.axis_label = 'Date'
fig.yaxis.axis_label = '腸病毒人數'
future_4_weeks = []
data_dict = {}
model_cds = {}
pred_4week = pd.read_csv('prediction_4week.csv')
pred_64week = pd.read_csv('prediction_64week.csv')
pred_4week_mae = pd.read_csv('mae_4week.csv')
pred_64week_mae = pd.read_csv('mae_64week.csv')
for area in areas:
for model in models:
data_dict['{}_{}_4week'.format(area,
model)] = pred_4week['{}_{}'.format(
area, model)].values
data_dict['{}_{}_64week'.format(
area, model)] = pred_64week['{}_{}'.format(area, model)].values
data_dict['{}_{}_4week_mae'.format(area, model)] = pred_4week_mae
data_dict['x_4week'] = pd.to_datetime(pred_4week['x'].values)
data_dict['x_64week'] = pd.to_datetime(pred_64week['x'].values)
dates = data_dict['x_64week']
fig_predict = figure(title="predictions",
tools=TOOLS,
width=600,
height=300,
x_axis_type='datetime')
mae_source_4week = ColumnDataSource(data=dict())
mae_source_64week = ColumnDataSource(data=dict())
columns = [
TableColumn(field="model", title="model Name"),
TableColumn(field="mae", title="MAE")
]
data_table_4week = DataTable(source=mae_source_4week,
columns=columns,
width=800)
data_table_64week = DataTable(source=mae_source_64week,
columns=columns,
width=800)
stats_4week = PreText(text='', width=500)
stats_64week = PreText(text='', width=500)
fig_predict_1year = figure(title="predictions_1year",
tools=TOOLS,
width=1000,
height=300,
x_axis_type='datetime',
x_axis_location="above",
background_fill_color="#efefef",
x_range=(dates[0], dates[20]))
select = figure(
title=
"Drag the middle and edges of the selection box to change the range above",
plot_height=130,
plot_width=1000,
y_range=fig_predict_1year.y_range,
x_axis_type="datetime",
y_axis_type=None,
tools="",
toolbar_location=None,
background_fill_color="#efefef")
range_rool = RangeTool(x_range=fig_predict_1year.x_range)
range_rool.overlay.fill_color = "navy"
range_rool.overlay.fill_alpha = 0.2
for i, model in enumerate(models):
model_cds[model] = {}
model_cds[model]['cds_4week'] = ColumnDataSource(data={
'x': [],
'y': []
})
model_cds[model]['cds_64week'] = ColumnDataSource(data={
'x': [],
'y': []
})
fig_predict.line(x='x',
y='y',
source=model_cds[model]['cds_4week'],
line_color=color_list[i],
muted_color=color_list[i],
line_width=2,
line_alpha=0.7,
legend=model)
fig_predict_1year.line(x='x',
y='y',
source=model_cds[model]['cds_64week'],
line_color=color_list[i],
muted_color=color_list[i],
line_width=2,
line_alpha=0.7,
legend=model)
select.line('x',
'y',
source=model_cds[model]['cds_64week'],
line_color=color_list[i],
muted_color=color_list[i])
select.ygrid.grid_line_color = None
select.add_tools(range_rool)
select.toolbar.active_multi = range_rool
fig_config(fig_predict)
fig_config(fig_predict_1year)
p = figure(title="Taiwan",
tools=TOOLS,
x_axis_location=None,
y_axis_location=None,
width=600,
height=800)
p.background_fill_color = 'beige'
p.border_fill_color = 'black'
p.border_fill_alpha = 0.05
p.grid.grid_line_color = None
p.toolbar.logo = None
p.toolbar_location = None
for i in range(22):
twgeo_json['features'][i]['properties']['line_width'] = 0.5
twgeo_json['features'][i]['properties']['line_color'] = 'white'
geo_source.geojson = json.dumps(twgeo_json)
selected_g = Patches(fill_color='blue',
line_color='white',
line_width=4,
line_alpha=0.5)
renderer = p.patches('xs',
'ys',
fill_alpha=0.7,
fill_color='blue',
line_color='line_color',
line_width='line_width',
line_alpha=1,
source=geo_source)
renderer.selection_glyph = selected_g
p.add_tools(TapTool())
p.add_tools(
HoverTool(tooltips=[("name",
"@name"), ("index",
"$index"), ("(Lon,Lat)", "($x, $y)")]))
def my_tap_handler(attr, old, new):
fig_config(fig_predict)
idx = geo_source.selected.indices[0]
area = twgeo_json['features'][idx]['properties']['area']
geo_source.selected.indices = area_dict[area]['area_idx']
for i in range(22):
if twgeo_json['features'][i]['properties']['area'] == area:
twgeo_json['features'][i]['properties']['line_width'] = 4
twgeo_json['features'][i]['properties']['line_color'] = 'black'
else:
twgeo_json['features'][i]['properties']['line_width'] = 0.5
twgeo_json['features'][i]['properties']['line_color'] = 'white'
geo_source.geojson = json.dumps(twgeo_json)
if area:
temp_date_4week = data_dict['x_4week']
temp_date_64week = data_dict['x_64week']
temp_text_4week = ''
for i, model in enumerate(models):
temp_y_4week = data_dict['{}_{}_4week'.format(area, model)]
temp_y_64week = data_dict['{}_{}_64week'.format(area, model)]
model_cds[model]['cds_4week'].data = {
'x': temp_date_4week,
'y': temp_y_4week
}
model_cds[model]['cds_64week'].data = {
'x': temp_date_64week,
'y': temp_y_64week
}
stats_4week.text = area + '(last 4 weeks mae):\n'
stats_64week.text = area + '(last one year mae):\n'
mae_source_4week.data = {
'model': [model for model in models if model != 'truth'],
'mae': [
pred_4week_mae['{}_{}'.format(area, model)]
for model in models if model != 'truth'
],
}
mae_source_64week.data = {
'model': [model for model in models if model != 'truth'],
'mae': [
pred_64week_mae['{}_{}'.format(area, model)]
for model in models if model != 'truth'
],
}
geo_source.on_change('selected', my_tap_handler)
layouts = column(fig_predict, stats_4week, data_table_4week)
layouts = row(p, layouts)
layouts = column(layouts, fig_predict_1year, select, stats_64week,
data_table_64week)
return layouts
