python2 python3 TypeError: unsupported operand type(s) for /: 'dict_values' and 'int'

plot绘制各种曲线的程序，在python27下运行正常，到python35下 提示出现这个错误：

TypeError: unsupported operand type(s) for /: 'dict_values' and 'int'

 

• In python3, dict.values returns a dict_values object, which is not a list or tuple. Try coercing that into a list. total_minutes = list(total_minutes_by_account.values()). – Abdou Apr 27 '17 at 16:40

• It's a good idea to check the results of a np.array(...) statement. During testing look at total_minutes, or at least check its shape and dtype. Don't just assume, check. In this case the dtype is probably object, not floats or ints. – hpaulj Apr 27 '17 at 17:29

• Find below a summary of his answer. It helped me a lot.

  In [1618]: dd = {'a':[1,2,3], 'b':[4,5,6]}
  In [1619]: dd
  Out[1619]: {'a': [1, 2, 3], 'b': [4, 5, 6]}
  In [1620]: dd.values()
  Out[1620]: dict_values([[1, 2, 3], [4, 5, 6]])
  In [1621]: np.mean(dd.values())
  ... 
  TypeError: unsupported operand type(s) for /: 'dict_values' and 'int'
  

  Solution: convert the dict_values to list:

  In [1623]: list(dd.values())
  Out[1623]: [[1, 2, 3], [4, 5, 6]]
  In [1624]: np.mean(list(dd.values()))
  Out[1624]: 3.5
  

  In Py3, range and dict.keys() require the same extra touch.

  ========

  np.mean first tries to convert the input to an array, but with values() that isn't what we want. It makes a single item object array containing this whole object.

• In [1626]: np.array(dd.values())
  Out[1626]: array(dict_values([[1, 2, 3], [4, 5, 6]]), dtype=object)
  In [1627]: _.shape
  Out[1627]: ()
  In [1628]: np.array(list(dd.values()))
  Out[1628]: 
  array([[1, 2, 3],
         [4, 5, 6]])

  25行修改后 28行也要加list函数 ，否则提示如下错误（下面代码黑色两行，原来python2中无list函数）

• numpy.core._internal.AxisError: axis -1 is out of bounds for array of dimension 0

• import numpy as np

  import matplotlib
  matplotlib.use('Agg')
  import matplotlib.pyplot as plt

  import collections
  import time
  import pickle as pickle

  _since_beginning = collections.defaultdict(lambda: {})
  _since_last_flush = collections.defaultdict(lambda: {})

  _iter = [0]
  def tick():
      _iter[0] += 1

  def plot(name, value):
      _since_last_flush[name][_iter[0]] = value

  def flush():
      prints = []

      for name, vals in _since_last_flush.items():
          prints.append("{}\t{}".format(name, np.mean(list(vals.values()))))
          _since_beginning[name].update(vals)
          
          x_vals = np.sort(list(_since_beginning[name].keys()))
          y_vals = [_since_beginning[name][x] for x in x_vals]

          plt.clf()
          plt.plot(x_vals, y_vals)
          plt.xlabel('iteration')
          plt.ylabel(name)
          plt.savefig(name.replace(' ', '_')+'.jpg')

      print ("iter {}\t{}".format(_iter[0], "\t".join(prints)))
      _since_last_flush.clear()

      with open('log.pkl', 'wb') as f:
          pickle.dump(dict(_since_beginning), f, pickle.HIGHEST_PROTOCOL)