def generate_delta_report(df,
delta_cols,
target_col,
title='Delta report',
save_to='unnamed_delta_report.html'):
p1 = plot_cat_bokeh(df[delta_cols + [target_col]], title=title)
p1.title.text_font_size = '18pt'
# deltas
df_delta = pd.DataFrame(index=df.index)
df_drel = pd.DataFrame(index=df.index)
for col in delta_cols:
df_delta['delta_' + col] = df[col] - df[target_col]
df_drel['pct_delta_' +
col] = 100 * (df[col] - df[target_col]) / df[target_col]
p2 = plot_cat_bokeh(df_delta,
title='Delta analysis vs ' + target_col,
color_offset=0)
p3 = plot_cat_bokeh(df_drel,
title='Percent delta vs ' + target_col,
color_offset=0)
# histogram
p4 = plot_hist_bokeh(df_delta, bins=100, color_offset=0)
# box plot
p5 = plot_boxplot_bokeh(df_delta)
p6 = plot_acf_bokeh(df_delta)
p = column(p1, p2, p3, Row(p4, p5), p6)
if save_to is None:
return p
else:
output_file(save_to, title=title)
save(p)
def plot_bar_bokeh_from_dict(dict_df,
title='Bar plot',
plot_width=800,
plot_height=300):
p = [
plot_bar_bokeh(dict_df[k],
title=title + ' ' + k,
plot_height=plot_height,
plot_width=plot_width) for k in dict_df.keys()
]
return Row(*p)
def generate_error_report(df,
title='Error report',
x_type='event_date',
save_to='unnamed_error_report.html'):
# normal is already deltas
p1 = plot_cat_bokeh(df.loc[:, df.count() > 0], title=title, x_type=x_type)
p1.title.text_font_size = '18pt'
# histogram
p2 = plot_hist_bokeh(df, bins=100, color_offset=0)
# box plot
p3 = plot_boxplot_bokeh(df)
p4 = plot_acf_bokeh(df)
p = Column(p1, Row(p2, p3), p4)
if save_to is None:
return p
else:
output_file(save_to, title=title)
save(p)
def plot_scatter_bokeh_size_vars(df,
x,
y,
c,
size_vars,
title="Scatter plot",
save_to="unnamed_scatter_plot.html"):
output_file(save_to, title=title)
f = []
for sv in size_vars:
f.append(
plot_scatter_bokeh(df,
x,
y,
c,
sv,
title=title + " - sized by " + sv))
p = Row(*f)
save(p)
d5['end_datetime_str'].push( d6['end_datetime_str'][draw_index]);
d5['start_datetime_dt'].push( d6['start_datetime_dt'][draw_index]);
d5['end_datetime_dt'].push( d6['end_datetime_dt'][draw_index]);
source_sx_draw_highlight.data = d5;
source_sx_draw_highlight.change.emit();
""")
silder_draw_index.js_on_change('value', callback)
columns = [
TableColumn(field="start_datetime_dt", title="start_datetime_dt"),
TableColumn(field="sx_value", title="sx_value")
]
data_table = DataTable(source=source_sx_persistency, columns=columns, width=400, height=280)
controles=Column(silder_draw_index, slider_sx, sizing_mode='stretch_width')
p_sx_layout = Column(p_sx, controles, margin=( 8 , 8 , 8 , 8 ))
toggles = Row(toggle1, toggle2)
p_crd_layout = Column(p_crd, toggles, slider_alpha_Value, data_table, margin=( 8 , 8 , 8 , 8 ))
#最後の部分
plots = Row(p_sx_layout, p_crd_layout)
#output_file("MOGE.html")
show(plots)
curdoc().add_root(plots)
def showGUI(self, pth_to_img, y_form, pred):
'''
Method builds the bokeh GUI
Parameters
----------
pth_to_img: path to ultrasound image
y_form: true form of the lesion
pred: predicted form the lesion
'''
##############
#Set up a figure
##############
p = figure(x_range=(0, self.DIMS[0]),
y_range=(0, self.DIMS[1]),
tools=self._tools_to_show,
plot_width=self.DIMS[0],
plot_height=self.DIMS[1],
toolbar_location="above")
#Add image as background
p.image_url(url=[self.root_pth + pth_to_img],
x=431,
y=302,
w=862,
h=604,
anchor="center")
#Nicier plot
self._makeShiny(plot=p)
##############
#Add lines and plot them
##############
src_true, src_pred = self._getData()
self._plotLines(plot=p, src_true=src_true, src_pred=src_pred)
##############
#Add table
##############
table = self._addTable(src_pred=src_pred)
##############
#Add polygons
##############
true_pol, c_t = self._addLesionForm(form=y_form, color='red', plot=p)
pred_pol, c_p = self._addLesionForm(form=pred, color='blue', plot=p)
#Add toggles for polygons
toggle_true = Toggle(label="Show true form",
button_type="primary",
active=True)
toggle_true.js_link('active', true_pol, 'visible')
toggle_true.js_link('active', c_t, 'visible')
toggle_pred = Toggle(label="Show predicted form",
button_type="primary",
active=True)
toggle_pred.js_link('active', pred_pol, 'visible')
toggle_true.js_link('active', c_p, 'visible')
##############
#Add download button
##############
button_csv = Button(label="Download", button_type="primary")
button_csv.callback = CustomJS(args=dict(source=src_pred),
code=open(self.root_pth +
"download.js").read())
##############
#Add title div
##############
div_title = Div(text="""<div> <b>LESION ADJUSTER</b> </div>""",
align='center',
style={
'font-size': '150%',
'color': '#1f77b4'
})
##############
#Add description to the buttons
##############
div_desc = Div(text="""<div> <b>CONTROLS</b> </div>""",
align='center',
style={
'font-size': '110%',
'color': '#1f77b4'
})
##############
#Add Div to show euclidean distance and button to recalculate it
##############
div_euclid = Div(text="""
<b>Diameter of predicted form is:</b> 334.80 <br>
<b>Diameter of true form is:</b> 368.64 <br>
<b>RMSE is:</b> 34.13
""",
align='center',
style={'font-size': '100%'})
p.js_on_event(
events.MouseMove,
CustomJS(args=dict(div=div_euclid,
source_data_pred=src_pred,
source_data_true=src_true),
code="""
var data_p = source_data_pred.data;
var data_t = source_data_true.data;
var x_p = data_p['x']
var y_p = data_p['y']
var x_t = data_t['x']
var y_t = data_t['y']
var diam_p = 0
var diam_t = 0
var rmse = 0
//Diameter of pred form
diam_p = Math.sqrt(Math.pow((x_p[0]-x_p[1]),2) + Math.pow((y_p[0]-y_p[1]),2))
//Diameter of true form
diam_t = Math.sqrt(Math.pow((x_t[0]-x_t[1]),2) + Math.pow((y_t[0]-y_t[1]),2))
//RMSE
rmse = Math.sqrt(Math.pow(diam_p - diam_t,2)/1)
//Result
div.text = "<b>Diameter of predicted form is: </b>" + diam_p.toFixed(2) + "<br> <b>Diameter of true form is: </b>" + diam_t.toFixed(2) + " <br> <b>RMSE is: </b>" + rmse.toFixed(2);
"""))
##############
#Show
##############
show(
Column(
div_title,
Row(
Column(p, table),
Column(div_desc, toggle_true, toggle_pred, button_csv,
div_euclid))))
def main():
# set up main bokeh figure
p = figure(x_range=(0, 10),
y_range=(0, 10),
tools=[],
title='Draw points in the network')
p.background_fill_color = 'lightgrey'
# start off with sample points and their associated flows
source = ColumnDataSource({
'x': [2, 7, 5, 8],
'y': [2, 2, 6, 1],
'flow': ['-2', '-5', '8', '-1']
})
renderer = p.scatter(x='x', y='y', source=source, color='blue', size=10)
columns = [
TableColumn(field="x", title="x"),
TableColumn(field="y", title="y"),
TableColumn(field='flow', title='flow')
]
table = DataTable(source=source,
columns=columns,
editable=True,
height=200)
draw_tool = PointDrawTool(renderers=[renderer], empty_value='1')
p.add_tools(draw_tool)
p.toolbar.active_tap = draw_tool
titletextbox = Div(
text=
"<h2>Objective: minimize construction cost of network<p>Construction cost is based on number of pipes and distance between nodes.<br>Additional constraints imposed: flows in network must be balanced.<br></h2>",
width=1100,
height=150)
textbox = Div(text="", width=200, height=100)
floating = 1.
fixed = 0.
button = Button(label='Solve Network')
button.on_event(
ButtonClick,
partial(button_click_event, source=source, textbox=textbox, figure=p))
p.on_event(
PanEnd,
partial(button_click_event, source=source, textbox=textbox, figure=p))
# set up sliders
lumpSumCost = Slider(title="Fixed cost pipe",
value=0.0,
start=0.0,
end=500.0,
step=50)
floatingCost = Slider(title="Floating cost pipe",
value=1.0,
start=0.0,
end=500.0,
step=10.)
for w in [lumpSumCost, floatingCost]:
w.on_change(
'value',
partial(update_data,
source=source,
textbox=textbox,
figure=p,
lumpSumCost=lumpSumCost,
floatingCost=floatingCost))
# create page layout
curdoc().add_root(
Column(
titletextbox,
Row(Column(p, table, width=800),
Column(lumpSumCost, floatingCost, button, textbox,
width=300))))
curdoc().title = "Network"
# print db.get(str(id)+"4"+"x")
# print db.get(str(id)+"5"+"x")
# print db.get(str(id)+"1"+"y")
# print db.get(str(id)+"2"+"y")
# print db.get(str(id)+"3"+"y")
# print db.get(str(id)+"4"+"y")
# print db.get(str(id)+"5"+"y")
# print db.getall()
#db.deldb()
#db.dump()
#db.deldb()
#db.dump()
layout = (Column(p, Row(button1, button)))
tools = ["xpan,pan,xwheel_zoom,wheel_zoom,box_zoom,reset,previewsave"]
q = figure(x_range=(0, 10),
y_range=(0, 10),
tools=tools,
title='Analysis Peak')
q.y_range.start = -10
q.y_range.end = 220
q.x_range.start = -10
q.x_range.end = 180
q.xaxis.axis_label = "Extension [nm]"
q.xaxis.axis_label_text_font_size = "20pt"
q.xaxis.axis_label_text_font_style = 'bold'
def new_scatter():
df = pd.read_pickle("./data/Rat_Sightings.pkl")
def var_var_scatter(df=df,
var1="Borough",
var2="Neighborhood",
year="All",
season="All"):
"""
Allows the user to select whichever variables
from the dataset for comparison in the form
of a scatter plot.
"""
# we want the var1 to be the least granular variable
name1, name2 = var1, var2
# get the number of classes
n_classes1 = df[var1].nunique()
n_classes2 = df[var2].nunique()
# if the num of classes in var2 is less then reassign variables
if n_classes2 < n_classes1:
var1 = name2
var2 = name1
# group by
if year == "All" and season == "All":
#group dataframe by indicated column name and rename columns
groupeddf = (df.assign(n=0).groupby([
var1, var2
]).n.count().reset_index().rename(columns={"n": "rat_sightings"}))
elif year != "All" and season == "All":
#group dataframe by indicated column name and rename columns
groupeddf = (df.query("Year == '%s'" % year).assign(n=0).groupby([
var1, var2
]).n.count().reset_index().rename(columns={"n": "rat_sightings"}))
elif year != "All" and season != "All":
#group dataframe by indicated column name and rename columns
groupeddf = (df.query("Year == '%s'" % year).query(
"Season == '%s'" % season).assign(n=0).groupby(
[var1, var2]).n.count().reset_index().rename(
columns={"n": "rat_sightings"}))
elif year == "All" and season != "All":
#group dataframe by indicated column name and rename columns
groupeddf = (df.query("Season == '%s'" % season).assign(
n=0).groupby([var1, var2]).n.count().reset_index().rename(
columns={"n": "rat_sightings"}))
else:
#group dataframe by indicated column name and rename columns
groupeddf = (df.assign(n=0).groupby([
var1, var2
]).n.count().reset_index().rename(columns={"n": "rat_sightings"}))
# get unique values for both vars
uniq_vals1 = [*groupeddf[var1].unique()]
uniq_vals2 = [*groupeddf[var2].unique()]
# now add in percentages to give more information to user
# get the total values
sum_data = pd.DataFrame(groupeddf.groupby(var1)["rat_sightings"].sum())
col_name = "total_%s_sights" % var1
sum_data.columns = [col_name]
# merge the totals in to the dataframe
""" Sure this can be applied to the dataframe in some fashion but opting
for the conceptually easier solution here.
"""
groupeddf = groupeddf.merge(sum_data.reset_index())
# get percentages for each location type
perc_name = "perc_sights"
groupeddf[perc_name] = (groupeddf["rat_sightings"] /
groupeddf[col_name]) * 100
# create color mapper
mapper = LogColorMapper(palette=Blues9[::-1])
# instantiate figure
p = figure(plot_width=800, plot_height=600, y_range=uniq_vals1)
p.circle(x="rat_sightings",
y=jitter(var1, width=0.6, range=p.y_range),
source=ColumnDataSource(data=groupeddf),
alpha=0.6,
size=30,
hover_alpha=0.9,
fill_color={
"field": "rat_sightings",
"transform": mapper
})
p.add_tools(
HoverTool(tooltips=[(
"%s" % var1, "@%s" %
var1), ("%s" % var2, "@%s" %
var2), ("Num. of Rat Sightings", "@rat_sightings"),
("Percentage of Rat Sightings within the %ss" %
var1, "@perc_sights %")]))
return p
# define variable selects
var1_select = Select(value="Borough",
options=[*df.columns],
title="Select a variable: ")
var2_select = Select(value="Neighborhood",
options=[*df.columns],
title="Select a variable: ")
year_select = Select(value="All",
options=[*df["Year"].unique()] + ["All"],
title="Select a year to view: ")
season_select = Select(value="All",
options=[*df["Season"].unique()] + ["All"],
title="Select a season to view: ")
# define interactivity
def update_plot(attr, old, new):
layout.children[1] = var_var_scatter(var1=var1_select.value,
var2=var2_select.value,
year=year_select.value,
season=season_select.value)
# inc interactivity
var1_select.on_change("value", update_plot)
var2_select.on_change("value", update_plot)
year_select.on_change("value", update_plot)
season_select.on_change("value", update_plot)
layout = Column(
Row(Column(var1_select, var2_select), Column(year_select,
season_select)),
var_var_scatter())
return layout