def createBokehChart(self):
plotwidth = int(self.getPreferredOutputWidth() / 2)
plotheight = int(self.getPreferredOutputHeight() * 0.75)
defaultbin = math.sqrt(len(self.getWorkingPandasDataFrame().index))
binsize = int(self.options.get('binsize', defaultbin))
valueFields = self.getValueFields()
histograms = []
if len(valueFields) != 1:
for i, valueField in enumerate(valueFields):
color = self.options.get("color")
if color is None:
color = Colors.hexRGB(1. * i / 2)
histograms.append(
Histogram(self.getWorkingPandasDataFrame(),
values=valueField,
plot_width=plotwidth,
plot_height=plotheight,
bins=binsize,
color=color,
xgrid=True,
ygrid=True,
ylabel='Frequency'))
return histograms
else:
return Histogram(self.getWorkingPandasDataFrame(),
values=self.getValueFields()[0],
bins=binsize,
color=self.options.get("color"),
xgrid=True,
ygrid=True,
ylabel='Frequency')
def matplotlibRender(self, fig, ax):
stacked = len(self.getValueFields()) > 1 and self.options.get(
"histoChartType", "stacked") == "stacked"
subplots = self.isSubplots()
defaultbin = math.sqrt(len(self.getWorkingPandasDataFrame().index))
binsize = int(self.options.get('binsize', defaultbin))
def plot(ax, valueField=None, color=None):
data = self.getWorkingPandasDataFrame(
) if valueField is None else self.getWorkingPandasDataFrame(
)[valueField]
data.plot(kind="hist",
stacked=stacked,
ax=ax,
bins=binsize,
legend=self.showLegend(),
x=valueField,
label=valueField,
color=color,
colormap=None if color else Colors.colormap)
if subplots:
for j, valueField in enumerate(self.getValueFields()):
plot(self.getAxItem(ax, j), valueField,
Colors.colormap(1. * j / 2))
else:
plot(ax)
def matplotlibRender(self, fig, ax):
stacked = len(self.getValueFields()) > 1 and self.options.get("histoChartType", "stacked") == "stacked"
subplots = self.isSubplots()
defaultbin = math.sqrt(len(self.getWorkingPandasDataFrame().index))
binsize = int(self.options.get('binsize', defaultbin))
def plot(ax, valueField=None, color=None):
data = self.getWorkingPandasDataFrame() if valueField is None else self.getWorkingPandasDataFrame()[valueField]
data.plot(
kind="hist", stacked=stacked, ax=ax, bins=binsize, legend=self.showLegend(), x = valueField,
label=valueField,color = color, colormap = None if color else Colors.colormap
)
if subplots:
for j, valueField in enumerate(self.getValueFields()):
plot(self.getAxItem(ax, j), valueField, Colors.colormap(1.*j/2))
else:
plot(ax)
def matplotlibRender(self, fig, ax):
rug=self.options.get("rug","false") == "true"
kde=self.options.get("kde","true") == "true"
defaultbin = math.sqrt(len(self.getWorkingPandasDataFrame().index))
binsize = int(self.options.get('binsize', defaultbin))
def plot(ax, valueField=None, color=None):
data = self.getWorkingPandasDataFrame()[valueField]
sns.distplot( data, ax=ax, bins=binsize, rug=rug, kde=kde,
kde_kws={"label":"{0} KDE Estim".format(valueField)}, hist_kws={"label":"{0} Freq".format(valueField)},
label=valueField,color = color
)
if len(self.getValueFields()) > 1:
for j, valueField in enumerate(self.getValueFields()):
plot(self.getAxItem(ax, j), valueField, Colors.colormap(1.*j/2))
else:
plot(ax, self.getValueFields()[0])
def createBokehChart(self):
keyFields = self.getKeyFields()
valueFields = self.getValueFields()
clusterby = self.options.get("clusterby")
subplots = self.isSubplot()
workingPDF = self.getWorkingPandasDataFrame().copy()
for index, row in workingPDF.iterrows():
for k in keyFields:
if isinstance(row[k],
str if sys.version >= '3' else basestring):
row[k] = row[k].replace(':', '.')
workingPDF.loc[index] = row
charts = []
if clusterby is not None and (subplots or len(valueFields) <= 1):
subplots = subplots if len(valueFields) == 1 or subplots else False
for j, valueField in enumerate(valueFields):
pivot = workingPDF.pivot(index=keyFields[0],
columns=clusterby,
values=valueField)
if not subplots:
fig = figure(
x_range=self.safeList(pivot.index) if not np.
issubdtype(pivot.index.dtype, np.number) else None)
charts.append(fig)
for i, col in enumerate(pivot.columns[:10]): #max 10
if subplots:
charts.append(
Line(pivot[col].values,
color=Colors.hexRGB(1. * i / 2),
ylabel=valueField,
xlabel=keyFields[0],
legend=False,
title="{0} = {1}".format(
clusterby, pivot.columns[i])))
else:
xValues = pivot.index.values.tolist()
if not np.issubdtype(pivot.index.dtype, np.number):
xValues = range(1, len(xValues) + 1)
fig.line(
x=xValues,
y=pivot[col].values,
color=Colors.hexRGB(1. * i / 2),
legend=str(col) if self.showLegend() else None)
else:
if subplots:
for i, valueField in enumerate(valueFields):
charts.append(
Line(workingPDF,
x=keyFields[0],
y=valueField,
color=Colors.hexRGB(1. * i / 2),
legend=self.showLegend(),
plot_width=int(800 / len(valueFields))))
else:
charts.append(
Line(workingPDF,
x=keyFields[0],
y=valueFields,
color=valueFields,
legend=self.showLegend()))
if clusterby is not None:
self.addMessage(
"Warning: 'Cluster By' ignored when grouped option with multiple Value Fields is selected"
)
return charts
def createBokehChart(self):
keyFields = self.getKeyFields()
valueFields = self.getValueFields()
clusterby = self.options.get("clusterby")
stacked = self.options.get("charttype", "grouped") == "stacked"
subplots = self.isSubplot()
workingPDF = self.getWorkingPandasDataFrame().copy()
def convertPDFDate(df, col):
#Bokeh doesn't support datetime as index in Bar chart. Convert to String
if len(keyFields) == 1:
dtype = df[col].dtype.type if col in df else None
if numpy.issubdtype(dtype, numpy.datetime64):
dateFormat = self.options.get("dateFormat", None)
try:
df[col] = df[col].apply(lambda x: str(x).replace(':','-') if dateFormat is None else x.strftime(dateFormat))
except:
self.exception("Error converting dateFormat {}".format(dateFormat))
df[col] = df[col].apply(lambda x: str(x).replace(':','-'))
for index, row in workingPDF.iterrows():
for k in keyFields:
if isinstance(row[k], str if sys.version >= '3' else basestring):
row[k] = row[k].replace(':', '.')
workingPDF.loc[index] = row
charts=[]
def goChart(label, stack_or_group, values, ylabel=None, color=None):
convertPDFDate(workingPDF, keyFields[0])
if ylabel is None:
ylabel=values
label=label if isinstance(label, (list, tuple)) else [label]
if stacked:
charts.append( Bar(workingPDF, label=CatAttr(columns=label, sort=False), stack=stack_or_group, color=color, values=values, legend=self.showLegend(), ylabel=ylabel))
else:
charts.append( Bar(workingPDF, label=CatAttr(columns=label, sort=False), group=stack_or_group, color=color, values=values, legend=self.showLegend(), ylabel=ylabel))
if clusterby is not None and (subplots or len(valueFields)<=1):
subplots = subplots if len(valueFields)==1 or subplots else False
if subplots:
for j, valueField in enumerate(valueFields):
pivot = workingPDF.pivot(
index=keyFields[0], columns=clusterby, values=valueField
)
for i,col in enumerate(pivot.columns[:10]): #max 10
data = pd.DataFrame({'values':pivot[col].values, 'names': pivot.index.values})
convertPDFDate(data, 'names')
if subplots:
charts.append(
Bar(data, label=CatAttr(columns=['names'], sort=False), color = Colors.hexRGB( 1.*i/2 ), values='values', ylabel=valueField, legend=False,
title="{0} = {1}".format(clusterby, pivot.columns[i])
)
)
else:
goChart( keyFields[0], clusterby, valueFields[0])
else:
if subplots:
for i,valueField in enumerate(valueFields):
goChart( keyFields[0], None, valueField, color=Colors.hexRGB( 1.*i/2 ))
else:
if len(valueFields) > 1:
series = '_'.join(valueFields)
values = blend(*valueFields, name=series.replace('_', ','), labels_name=series)
else:
series = False
values = valueFields[0]
goChart(keyFields, series, values, ylabel=','.join(valueFields))
if clusterby is not None:
self.addMessage("Warning: 'Cluster By' ignored when grouped option with multiple Value Fields is selected")
return charts