或用于python计算的模块

在我在Tradewave.net的测试中，TA-lib总是赢:

import talib as ta
import numpy as np
import pandas as pd
import scipy
from scipy import signal
import time as t

PAIR = info.primary_pair
PERIOD = 30

def initialize():
    storage.reset()
    storage.elapsed = storage.get('elapsed', [0,0,0,0,0,0])

def cumsum_sma(array, period):
    ret = np.cumsum(array, dtype=float)
    ret[period:] = ret[period:] - ret[:-period]
    return ret[period - 1:] / period

def pandas_sma(array, period):
    return pd.rolling_mean(array, period)

def api_sma(array, period):
    # this method is native to Tradewave and does NOT return an array
    return (data[PAIR].ma(PERIOD))

def talib_sma(array, period):
    return ta.MA(array, period)

def convolve_sma(array, period):
    return np.convolve(array, np.ones((period,))/period, mode='valid')

def fftconvolve_sma(array, period):    
    return scipy.signal.fftconvolve(
        array, np.ones((period,))/period, mode='valid')    

def tick():

    close = data[PAIR].warmup_period('close')

    t1 = t.time()
    sma_api = api_sma(close, PERIOD)
    t2 = t.time()
    sma_cumsum = cumsum_sma(close, PERIOD)
    t3 = t.time()
    sma_pandas = pandas_sma(close, PERIOD)
    t4 = t.time()
    sma_talib = talib_sma(close, PERIOD)
    t5 = t.time()
    sma_convolve = convolve_sma(close, PERIOD)
    t6 = t.time()
    sma_fftconvolve = fftconvolve_sma(close, PERIOD)
    t7 = t.time()

    storage.elapsed[-1] = storage.elapsed[-1] + t2-t1
    storage.elapsed[-2] = storage.elapsed[-2] + t3-t2
    storage.elapsed[-3] = storage.elapsed[-3] + t4-t3
    storage.elapsed[-4] = storage.elapsed[-4] + t5-t4
    storage.elapsed[-5] = storage.elapsed[-5] + t6-t5    
    storage.elapsed[-6] = storage.elapsed[-6] + t7-t6        

    plot('sma_api', sma_api)  
    plot('sma_cumsum', sma_cumsum[-5])
    plot('sma_pandas', sma_pandas[-10])
    plot('sma_talib', sma_talib[-15])
    plot('sma_convolve', sma_convolve[-20])    
    plot('sma_fftconvolve', sma_fftconvolve[-25])

def stop():

    log('ticks....: %s' % info.max_ticks)

    log('api......: %.5f' % storage.elapsed[-1])
    log('cumsum...: %.5f' % storage.elapsed[-2])
    log('pandas...: %.5f' % storage.elapsed[-3])
    log('talib....: %.5f' % storage.elapsed[-4])
    log('convolve.: %.5f' % storage.elapsed[-5])    
    log('fft......: %.5f' % storage.elapsed[-6])

结果:

[2015-01-31 23:00:00] ticks....: 744
[2015-01-31 23:00:00] api......: 0.16445
[2015-01-31 23:00:00] cumsum...: 0.03189
[2015-01-31 23:00:00] pandas...: 0.03677
[2015-01-31 23:00:00] talib....: 0.00700  # <<< Winner!
[2015-01-31 23:00:00] convolve.: 0.04871
[2015-01-31 23:00:00] fft......: 0.22306

虽然这里有这个问题的解决方案，但请看看我的解决方案。这是非常简单和工作良好。

import numpy as np
dataset = np.asarray([1, 2, 3, 4, 5, 6, 7])
ma = list()
window = 3
for t in range(0, len(dataset)):
    if t+window <= len(dataset):
        indices = range(t, t+window)
        ma.append(np.average(np.take(dataset, indices)))
else:
    ma = np.asarray(ma)

高效的解决方案

卷积比直接的方法好得多，但(我猜)它使用FFT，因此相当慢。但是，下面的方法特别适用于计算运行平均值

def running_mean(x, N):
    cumsum = numpy.cumsum(numpy.insert(x, 0, 0)) 
    return (cumsum[N:] - cumsum[:-N]) / float(N)

要检查的代码

In[3]: x = numpy.random.random(100000)
In[4]: N = 1000
In[5]: %timeit result1 = numpy.convolve(x, numpy.ones((N,))/N, mode='valid')
10 loops, best of 3: 41.4 ms per loop
In[6]: %timeit result2 = running_mean(x, N)
1000 loops, best of 3: 1.04 ms per loop

注意numpy。allclose(result1, result2)为True，两个方法等价。 N越大，时间差异越大。

警告:虽然cumsum更快，但会增加浮点错误，这可能导致您的结果无效/不正确/不可接受

这里的评论指出了这个浮点错误问题，但我在回答中让它更明显。

# demonstrate loss of precision with only 100,000 points
np.random.seed(42)
x = np.random.randn(100000)+1e6
y1 = running_mean_convolve(x, 10)
y2 = running_mean_cumsum(x, 10)
assert np.allclose(y1, y2, rtol=1e-12, atol=0)

the more points you accumulate over the greater the floating point error (so 1e5 points is noticable, 1e6 points is more significant, more than 1e6 and you may want to resetting the accumulators) you can cheat by using np.longdouble but your floating point error still will get significant for relatively large number of points (around >1e5 but depends on your data) you can plot the error and see it increasing relatively fast the convolve solution is slower but does not have this floating point loss of precision the uniform_filter1d solution is faster than this cumsum solution AND does not have this floating point loss of precision

上面的一个答案中有一个mab的注释，它有这个方法。瓶颈有move_mean，这是一个简单的移动平均:

import numpy as np
import bottleneck as bn

a = np.arange(10) + np.random.random(10)

mva = bn.move_mean(a, window=2, min_count=1)

Min_count是一个很方便的参数，它可以取数组中该点的移动平均值。如果你不设置min_count，它将等于window，并且直到window points的所有内容都将是nan。

你可以用以下方法计算运行平均值:

import numpy as np

def runningMean(x, N):
    y = np.zeros((len(x),))
    for ctr in range(len(x)):
         y[ctr] = np.sum(x[ctr:(ctr+N)])
    return y/N

但是速度很慢。

幸运的是，numpy包含一个卷积函数，我们可以用它来加快速度。运行均值相当于将x与一个长度为N的向量进行卷积，其中所有元素都等于1/N。卷积的numpy实现包括起始瞬态，所以你必须删除前N-1点:

def runningMeanFast(x, N):
    return np.convolve(x, np.ones((N,))/N)[(N-1):]

在我的机器上，快速版本要快20-30倍，这取决于输入向量的长度和平均窗口的大小。

请注意，卷积确实包括一个“相同”模式，它似乎应该解决开始的瞬态问题，但它在开始和结束之间分割。

从其他答案来看，我不认为这是问题所要求的，但我需要保持一个不断增长的值列表的运行平均值。

因此，如果你想保持从某个地方(站点，测量设备等)获取的值的列表和最近n个值更新的平均值，你可以使用下面的代码，这将最大限度地减少添加新元素的工作:

class Running_Average(object):
    def __init__(self, buffer_size=10):
        """
        Create a new Running_Average object.

        This object allows the efficient calculation of the average of the last
        `buffer_size` numbers added to it.

        Examples
        --------
        >>> a = Running_Average(2)
        >>> a.add(1)
        >>> a.get()
        1.0
        >>> a.add(1)  # there are two 1 in buffer
        >>> a.get()
        1.0
        >>> a.add(2)  # there's a 1 and a 2 in the buffer
        >>> a.get()
        1.5
        >>> a.add(2)
        >>> a.get()  # now there's only two 2 in the buffer
        2.0
        """
        self._buffer_size = int(buffer_size)  # make sure it's an int
        self.reset()

    def add(self, new):
        """
        Add a new number to the buffer, or replaces the oldest one there.
        """
        new = float(new)  # make sure it's a float
        n = len(self._buffer)
        if n < self.buffer_size:  # still have to had numbers to the buffer.
            self._buffer.append(new)
            if self._average != self._average:  # ~ if isNaN().
                self._average = new  # no previous numbers, so it's new.
            else:
                self._average *= n  # so it's only the sum of numbers.
                self._average += new  # add new number.
                self._average /= (n+1)  # divide by new number of numbers.
        else:  # buffer full, replace oldest value.
            old = self._buffer[self._index]  # the previous oldest number.
            self._buffer[self._index] = new  # replace with new one.
            self._index += 1  # update the index and make sure it's...
            self._index %= self.buffer_size  # ... smaller than buffer_size.
            self._average -= old/self.buffer_size  # remove old one...
            self._average += new/self.buffer_size  # ...and add new one...
            # ... weighted by the number of elements.

    def __call__(self):
        """
        Return the moving average value, for the lazy ones who don't want
        to write .get .
        """
        return self._average

    def get(self):
        """
        Return the moving average value.
        """
        return self()

    def reset(self):
        """
        Reset the moving average.

        If for some reason you don't want to just create a new one.
        """
        self._buffer = []  # could use np.empty(self.buffer_size)...
        self._index = 0  # and use this to keep track of how many numbers.
        self._average = float('nan')  # could use np.NaN .

    def get_buffer_size(self):
        """
        Return current buffer_size.
        """
        return self._buffer_size

    def set_buffer_size(self, buffer_size):
        """
        >>> a = Running_Average(10)
        >>> for i in range(15):
        ...     a.add(i)
        ...
        >>> a()
        9.5
        >>> a._buffer  # should not access this!!
        [10.0, 11.0, 12.0, 13.0, 14.0, 5.0, 6.0, 7.0, 8.0, 9.0]

        Decreasing buffer size:
        >>> a.buffer_size = 6
        >>> a._buffer  # should not access this!!
        [9.0, 10.0, 11.0, 12.0, 13.0, 14.0]
        >>> a.buffer_size = 2
        >>> a._buffer
        [13.0, 14.0]

        Increasing buffer size:
        >>> a.buffer_size = 5
        Warning: no older data available!
        >>> a._buffer
        [13.0, 14.0]

        Keeping buffer size:
        >>> a = Running_Average(10)
        >>> for i in range(15):
        ...     a.add(i)
        ...
        >>> a()
        9.5
        >>> a._buffer  # should not access this!!
        [10.0, 11.0, 12.0, 13.0, 14.0, 5.0, 6.0, 7.0, 8.0, 9.0]
        >>> a.buffer_size = 10  # reorders buffer!
        >>> a._buffer
        [5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, 12.0, 13.0, 14.0]
        """
        buffer_size = int(buffer_size)
        # order the buffer so index is zero again:
        new_buffer = self._buffer[self._index:]
        new_buffer.extend(self._buffer[:self._index])
        self._index = 0
        if self._buffer_size < buffer_size:
            print('Warning: no older data available!')  # should use Warnings!
        else:
            diff = self._buffer_size - buffer_size
            print(diff)
            new_buffer = new_buffer[diff:]
        self._buffer_size = buffer_size
        self._buffer = new_buffer

    buffer_size = property(get_buffer_size, set_buffer_size)

你可以测试它，例如:

def graph_test(N=200):
    import matplotlib.pyplot as plt
    values = list(range(N))
    values_average_calculator = Running_Average(N/2)
    values_averages = []
    for value in values:
        values_average_calculator.add(value)
        values_averages.append(values_average_calculator())
    fig, ax = plt.subplots(1, 1)
    ax.plot(values, label='values')
    ax.plot(values_averages, label='averages')
    ax.grid()
    ax.set_xlim(0, N)
    ax.set_ylim(0, N)
    fig.show()

这使: