python - 如何使用Bokeh正确创建HeatMap

Question

我正在尝试使用Bokeh而不是matplotlib复制this question中显示的HeatMap。我不能完全正确。现有的例子并没有帮助我理解我做错了什么。我谦虚的尝试

from bokeh.io import output_notebook; output_notebook()
from bokeh.charts import HeatMap, show
from bokeh.palettes import RdYlGn6
import pandas as pd
import numpy as np

nba = pd.read_csv(urlopen("http://datasets.flowingdata.com/ppg2008.csv"), index_col=0)

# Normalize the data columns and sort.
nba = (nba - nba.mean()) / (nba.max() - nba.min())
nba.sort('PTS', inplace=True)

score = []
for x in nba.apply(tuple):
  score.extend(x)

data = {
  'players': list(nba.index) * len(nba.columns),
  'metric':  list(nba.columns) * len(nba.index),
  'score':   score,
}

hm = HeatMap(data, x='metric', y='players',values='score', title='Fruits', stat=None)
show(hm)

给



请注意，尽管标题相似，但this不能回答我的问题。这不是一个相同的错误，我正在使用Bokeh 0.12.3

Answer 1

将数据metric的生成更改为按元素重复，它应该是正确的:

'metric': [item for item in list(nba.columns) for i in range(len(nba.index))],

因此，适用于我的代码如下:

from bokeh.charts import HeatMap, show, output_file
import pandas as pd, numpy as np
from urllib2 import urlopen

nba = pd.read_csv(urlopen("http://datasets.flowingdata.com/ppg2008.csv"), index_col=0)

# Normalize the data columns and sort.
nba = (nba - nba.mean()) / (nba.max() - nba.min())
nba.sort_values(by = 'PTS', inplace=True)

score = []
for x in nba.apply(tuple):
  score.extend(x)

data = {
  'players': list(nba.index) * len(nba.columns),
  'metric':  [item for item in list(nba.columns) for i in range(len(nba.index))],
  'score':   score,
}

output_file('test.html')
hm = HeatMap(data, x='metric', y='players',values='score', title='Fruits', stat=None)
show(hm)