Question

我有一个随机的时间序列数据，有四列，如：年，月，日，降水。我想计算不同法术长度的干/湿法术。我正在寻找一种更方便的方法来做到这一点，而目前正在做一些丑陋的代码，如下所示：

import numpy as np
data = np.loadtxt('Data Series.txt', usecols=(1,3))
dry = np.zeros(12)
wet = np.zeros(12)

rows,cols = data.shape #reading number of rows and columns into variables

for i in xrange (0,rows):
    for m in xrange(0,12):
        if data[i,1] == 0 and data[i-1,1] == 0 and data[i-2,1] == 0:
            if data[i,0] == m+1:
                dry[m] += 1.0
        if data[i,1] > 0 and data[i-1,1] > 0 and data[i-2,1] > 0:
            if data[i,0] == m+1:
                wet[m] += 1.0
print '3 Days Dry Spell\n', dry
print '3 Days Wet Spell\n', wet

现在，如果我想计算4,5,6天的法术，那么“如果数据[i，1] == 0和数据[i-1,1] == 0 ......”就变成了一个巨大的。任何人都可以帮助我，以便我可以只给出法术长度而不是这条长丑线吗？

Answer 1

你可能想尝试这样的事情：

# first extract precipitation data for later use
precipitation = [data[i][1] for i in xrange(0, rows)]

# then test the range (i, i+m)
all_dry = all([(data==0) for data in precipitation[i:i+m]])
all_wet = not any([(data==0) for data in precipitation[i:i+m]])
# of course you can also use
all_wet = all([(data>0) for data in precipitation[i:i+m]])

但请注意，此方法在测试相邻日期时会引入冗余计算，因此可能不适合处理大量数据。

<强>编辑：

好的，这一次让我们寻找一种更有效的方法。

# still extract precipitation data for later use first
precipitation = [data[i][1] for i in xrange(0, rows)]

# let's start our calculations by counting the longest consecutive dry days 
consecutive_dry = [1 if data == 0 else 0 for data in precipitation]
for i in xrange(1, len(consecutive_dry))
    if consecutive_dry[i] == 1:
        consecutive_dry[i] += consecutive_dry[i - 1]

# then you will see, if till day i there're m consecutive dry days, then:
consecutive_dry[i] >= m    # here is the test

# ...and it would be same for wet day testings.

这显然比上述方法更有效：为了测试具有M个连续范围的总共N天，前一个需要O（N * M）个运算来计算，这个需要O（N）。

重新编辑：

这是原始代码的编辑版本。由于您的代码可以运行，这也可以在您的PC上运行或者运行。

import numpy as np
data = np.loadtxt('Data Series.txt', usecols=(1,3))
dry = np.zeros(12)
wet = np.zeros(12)

rows,cols = data.shape #reading number of rows and columns into variables

# prepare 
precipitation = [data[i][1] for i in xrange(0, rows)]

# collecting data for consecutive dry days
consecutive_dry = [1 if data == 0 else 1 for data in precipitation]
for i in xrange(1, len(consecutive_dry))
    if consecutive_dry[i] == 1:
        consecutive_dry[i] += consecutive_dry[i - 1]

# ...and for wet days
consecutive_wet = [1 if data > 0 else 0 for data in precipitation]
for i in xrange(1, len(consecutive_wet))
    if consecutive_wet[i] == 1:
        consecutive_wet[i] += consecutive_wet[i - 1]

# set your day range here. 
day_range = 3

for i in xrange (0,rows):
    if consecutive_dry[i] >= day_range:
        month_id = data[i,0]
        dry[month_id - 1] += 1
    if consecutive_wet[i] >= day_range:
        month_id = data[i,0]
        wet[month_id - 1] += 1

print '3 Days Dry Spell\n', dry
print '3 Days Wet Spell\n', wet

请尝试这个，如果有任何问题请告诉我。

Answer 2

我发现以下方法可以方便地计算平均干燥期。我在这里编写代码，因为它可能对其他人有用：

import numpy as np
import itertools as itr

#Import daily rainfall time series#
rain_series = np.loadtxt('daily_rainfall_timeseries.txt')

#separate the group of zero values (dry days) in a list of lists#
d = [list(x[1]) for x in itr.groupby(rain_series, lambda x: x > 0) if not x[0]]

#Count the lengths of different dry spells#
d_len = [len(f) for f in d]

#Calculate the mean dry period#
mean_dry_spell = np.mean(d_len)

如何用Python计算干湿法术？

2 个答案: