为什么对于非常相似的数据帧，Pandas的绘图如此不同？

Question

下面的代码片段从print部分显示 执行时没有问题：

import pandas as pd
from   matplotlib import pyplot as plt
import numpy as np
import csv

ts = pd.Series(np.random.randn(1000), index=pd.date_range('1/1/2000', periods=1000))
df = pd.DataFrame(np.random.randn(1000, 3), index=ts.index, columns=list('ABC'))
print (df)
print (df.index)
print (df.columns)
#df = df.cumsum()   # this also is ok
df.plot()
plt.show()

                   A         B         C
2000-01-01  0.882544 -0.841398  1.745238
2000-01-02  1.798310  1.049662 -0.115292
2000-01-03  1.223243 -0.086322 -0.565204
2000-01-04 -0.587905 -0.609485  0.296248
2000-01-05 -1.603916 -0.397210  0.007550
2000-01-06 -0.821833  0.112760 -0.082558   
     ...
     ...
2002-09-22 -0.530537  0.373358  2.920919
2002-09-23  0.121657  0.634864 -0.964255
2002-09-24  1.153799  2.468507 -2.087136
2002-09-25 -1.079853  0.684926  1.556522
2002-09-26 -2.163454  0.874373  1.942925

[1000 rows x 3 columns]
DatetimeIndex(['2000-01-01', '2000-01-02', '2000-01-03', '2000-01-04',
               '2000-01-05', '2000-01-06', '2000-01-07', '2000-01-08',
               '2000-01-09', '2000-01-10',
               ...
               ...
               '2002-09-17', '2002-09-18', '2002-09-19', '2002-09-20',
               '2002-09-21', '2002-09-22', '2002-09-23', '2002-09-24',
               '2002-09-25', '2002-09-26'],
              dtype='datetime64[ns]', length=1000, freq='D')
Index(['A', 'B', 'C'], dtype='object')

并生成一个很好的绘图（如预期的那样）

但是，当我执行以下代码时

import datetime

        df = pd.read_csv(fullFileNameOutCSV, header=0, names=headerCSV,usecols=['SEK/EUR', 'SEK/DKK', 'SEK/NOK'])        
        pd.Timedelta(datetime.timedelta(days=1))
        df.index = pd.date_range('2000-01-01',periods=4888)
        df.index = pd.to_datetime(df.index)
        print (df)
        print (df.index)
        print (df.columns)   
        df.plot()
        plt.show() 

            SEK/EUR  SEK/DKK  SEK/NOK
2000-01-01   9.4696   7.4501   8.8550
2000-01-02   9.4025   7.4495   8.7745
2000-01-03   9.3050   7.4452   8.7335
2000-01-04   9.1800   7.4431   8.6295
2000-01-05   9.1650   7.4433   8.5900
2000-01-06   9.0985   7.4433   8.5585
     ...
     ...
2013-05-14   9.8188   7.4444   9.5858
2013-05-15   9.8005   7.4428   9.5655
2013-05-16   9.7823   7.4427   9.5548
2013-05-17   9.7825   7.4415   9.5628
2013-05-18   9.7645   7.4419   9.5620
2013-05-19   9.8030   7.4428   9.5705

[4888 rows x 3 columns]
DatetimeIndex(['2000-01-01', '2000-01-02', '2000-01-03', '2000-01-04',
               '2000-01-05', '2000-01-06', '2000-01-07', '2000-01-08',
               '2000-01-09', '2000-01-10',
               ...
               ...
               '2013-05-10', '2013-05-11', '2013-05-12', '2013-05-13',
               '2013-05-14', '2013-05-15', '2013-05-16', '2013-05-17',
               '2013-05-18', '2013-05-19'],
              dtype='datetime64[ns]', length=4888, freq='D')

情节不是我所期望的。 为了得到一个合理的绘图，我在这个代码段上尝试了许多不同的变体，但到目前为止，我还没有找到能够生成与第一个绘图类似的绘图的代码。注意，垂直轴刻度---是什么引起的？ 为什么我不能得到一个类似于第一个代码片段的图呢？你知道吗

Answer 1

在修改字符串以输出到csv文件的地方插入了以下代码，这很好-Eureka:-）

if '.' not in nums[3]:
    # Append '.00' to integer strings
    nums[3] = nums[3] + '.00'  

Answer 2

这是修改代码后的修正图

Answer 3

我使用下面的函数来计算每种货币的均值和方差。如果我一直在使用它，那么这3个异常值很容易被注意到-我现在正在使用它。你知道吗

def updateMeanVar(x,k,mu,vr):
    '''
     Purpose: Update the estimates for the mean and variance
              (recursive algorithm)              
       Inputs:
          x   new value (x_k)                 x_1,x_2,...
          k   counter (index) for new value   1,2,
         mu   previously estimated mean (x_k not included)
         vr   previously estimated variance (x_k not included)
       Otputs:
         mu   updated mean     (with x_k included)
         vr   updated variance (with x_k included)

      Refs.
        Donald E. Knuth. Seminumerical  Algorithms, volume 2 of The  Art  of     Computer  Programming,
        chapter 4.2.2, page 232. Addison-Wesley, Boston, third edition, 1998.
    '''
    delta = x - mu
    mu += delta/k
    vr += delta*(x - mu)  
    return mu,vr

以下是欧洲央行（ECB）货币汇率的部分产出，包括今天的估计。你知道吗

Processing ECB exchange rates ...
Saving data ...

ECB working days:  4889
strt date: 1999-01-04
stop date: 2018-02-02
Exchange rates stored in:
  Sweden  (SEK/EUR): E:/Data/ECB_ExchangeRates/Sweden/Rates.txt  (mean=  9.235,std=  0.496)
  Denmark (DKK/EUR): E:/Data/ECB_ExchangeRates/Denmark/Rates.txt (mean=  7.448,std=  0.011)
  Norway  (NOK/EUR): E:/Data/ECB_ExchangeRates/Norway/Rates.txt  (mean=  8.238,std=  0.565)
All rates stored in: E:/Data/ECB_ExchangeRates/AllRates.txt
                     E:/Data/ECB_ExchangeRates/AllRates.csv

以及总结图。