from .glueplotly import GluePlotly from plotly.graph_objects import FigureWidget from ipywidgets import IntText, Dropdown from IPython.display import clear_output import numpy as np from collections import deque import colorlover as cl class GlueSunburstPlotly (GluePlotly): default_size_marker = 3 focused_size_marker = 4 pca = None options = {} default_color = "Dataset" def __init__(self, data, dimensions, **kwargs): GluePlotly.__init__(self, data, dimensions, **kwargs) self.DefaultLayoutTitles('', '', '') self.options['grouping_limit'] = IntText(description = 'Group limiy', value = 12) self.options['grouping_limit'].observe(lambda v:self.updateRender(), names='value') cl_options = list(cl.scales['8']['qual'].keys()) cl_options.append(GlueSunburstPlotly.default_color) self.options['colorscale'] = Dropdown(description = 'Color Palette:', value = GlueSunburstPlotly.default_color, options = cl_options) self.options['colorscale'].observe(lambda v:self.updateRender(), names='value') self.DefaultLegend('v', 1.02, 1.0); self.updateRender() def createFigureWidget(self): data_lines = [] traces = [] i=0 if self.only_subsets == False: colors = [i for r in range(self.data.size)] colorscale = [[0,'#EEEEEE']] else: colors = [] colorscale = [] i=-1 values = [] labels = [] nodes_id = [] parents = [] self.masks = [] colors = [] df = self.data.to_dataframe() ids = 1 queue = deque() if self.options['colorscale'].value == GlueSunburstPlotly.default_color: color_set = [self.data.get_component(dim).color for dim in self.dimensions] else: color_set = cl.scales['8']['qual'][self.options['colorscale'].value] if (len(self.dimensions) > 8): color_set = cl.interp( color_set, len(dimensions) ) color_set = cl.to_rgb(color_set) color_set.insert(0,'rgb(255,255,255)') queue.append({'id':ids,'dimension':0, 'label':"Data", 'mask':[True for i in range(self.data.size)], 'parent':''}) while len(queue) > 0: toprocess = queue.popleft() mask = toprocess['mask'] id = toprocess['id'] label = toprocess['label'] value = np.count_nonzero(mask) if value > 0: labels.append(label) values.append(value) parents.append(toprocess['parent']) nodes_id.append(id) self.masks.append(mask) colors.append(color_set[toprocess['dimension']]) data = df[mask] if toprocess['dimension'] < len(self.dimensions): dimension = self.dimensions[toprocess['dimension']] dvalues = np.unique(data[dimension].ravel()) if hasattr(self.data[dimension].flatten(), 'codes'): for val in dvalues: ids += 1 process = {'id':ids,'dimension':toprocess['dimension']+1, 'label':str(val), 'parent':toprocess['id']} process['mask'] = (mask & (self.data[dimension] == val)) queue.append(process) elif len(dvalues) < self.options['grouping_limit'].value: for val in dvalues: ids += 1 process = {'id':ids,'dimension':toprocess['dimension']+1, 'label':str(val), 'parent':toprocess['id']} process['mask'] = (mask & (self.data[dimension] == val)) queue.append(process) else: hist, bin_edges = np.histogram(self.data[dimension].flatten(), bins='auto') if (len(bin_edges) > self.options['grouping_limit'].value): hist, bin_edges = np.histogram(self.data[dimension].flatten(), bins=self.options['grouping_limit'].value) for edge in range(len(bin_edges)-1): ids += 1 label = "{:.1f}".format(bin_edges[edge]) + " - " + "{:.1f}".format(bin_edges[edge+1]) process = {'id':ids,'dimension':toprocess['dimension']+1, 'label':label, 'parent':toprocess['id']} if edge == 0: process['mask'] = (mask & ((self.data[dimension] >= bin_edges[edge]) & (self.data[dimension] <= bin_edges[edge+1]))) else : process['mask'] = (mask & ((self.data[dimension] > bin_edges[edge]) & (self.data[dimension] <= bin_edges[edge+1]))) queue.append(process) trace = { 'type': "sunburst", 'ids': nodes_id, 'labels': labels, 'parents': parents, 'values': values, 'branchvalues' : 'total', 'maxdepth' : 4, 'outsidetextfont': { 'size': 20, 'color': "#377eb8" }, 'leaf': { 'opacity' : 1 }, 'marker': { 'line': { 'width': 2 }, 'colors' : colors }, 'domain':{ 'x': [0, 0.9], 'y': [0, 1], }, } traces.append(trace) layout = { 'title' : self.options['title'].value, 'margin' : { 'l':self.margins['left'].value, 'r':self.margins['right'].value, 'b':self.margins['bottom'].value, 't':self.margins['top'].value }, 'showlegend': True, 'xaxis': { 'title' : self.options['xaxis'].value, 'range' : [0,1], 'showgrid':False, 'showline':False, 'showticklabels':False, 'zeroline':False }, 'yaxis': { 'title' : self.options['yaxis'].value, 'range' : [0,1], 'showgrid':False, 'showline':False, 'showticklabels':False, 'zeroline':False }, 'showlegend': self.margins['showlegend'].value, 'legend' : { 'orientation' : self.margins['legend_orientation'].value, 'x' : self.margins['legend_xpos'].value, 'y' : self.margins['legend_ypos'].value } } for i in range(len(self.dimensions)): dimension = self.dimensions[i] trace = { 'type': "scatter", 'name' : dimension, 'textposition' : 'middle right', 'x' : [-1000], 'y' : [i], 'mode' : 'markers', 'marker': { 'color' : color_set[i+1], 'size' : 20, 'line': { 'width': 0, }, 'symbol' : 'square' } } traces.append(trace) return FigureWidget({ 'data': traces, 'layout': layout }) def updateRender(self): self.plotly_fig = self.createFigureWidget() self.updateCallbacks(); GluePlotly.display(self) def updateCallbacks(self): append = False if self.only_subsets == False: self.plotly_fig.data[0].on_click(lambda x,y,z : self.setSubset(x,y,z), append) append = True if self.on_selection_callback is not None: self.plotly_fig.data[0].on_click(self.on_selection_callback, append) def on_selection(self, callback): GluePlotly.on_selection(self, callback) self.updateCallbacks() def updateSelection(self, ids): #self.plotly_fig.data[0].update( # selectedpoints=ids, #) pass; def setSubset(self,trace,points,selector): if(self.parent is not None): mask = None for p in points.point_inds: if p > 0: if mask is None: mask = self.masks[p] else : mask = mask | self.masks[p+1] if mask is not None: point_inds = np.nonzero(mask)[0] self.parent.updateSelection(point_inds)