Download Quandl Data

Install Quandl

!pip install quandl -q

import pandas as pd
import quandl
import matplotlib.pyplot as plt
%matplotlib inline

Get Data and Save in Pandas DataFrame

If run into rate limits, register at Quandl and provide api_key in the get call.

import getpass

KEY = getpass.getpass()

df = quandl.get(
    "FRED/DEXCHUS", 
    start_date='2014-01-01', 
    end_date='2020-10-01',
    api_key=KEY
).rename(columns={'Value':'DEXCHUS'})
df.index.name = 'ds'

df

	DEXCHUS
ds
2014-01-02	6.0504
2014-01-03	6.0505
2014-01-06	6.0524
2014-01-07	6.0507
2014-01-08	6.0510
...	...
2020-09-25	6.8220
2020-09-28	6.8106
2020-09-29	6.8150
2020-09-30	6.7896
2020-10-01	6.7898

1690 rows × 1 columns

Plot Time Series Data

df.plot(figsize=(16, 10));

Find Missing Data

pd.date_range(
    start = df.index.min(), 
    end = df.index.max()
).difference(df.index)

DatetimeIndex(['2014-01-04', '2014-01-05', '2014-01-11', '2014-01-12',
               '2014-01-18', '2014-01-19', '2014-01-20', '2014-01-25',
               '2014-01-26', '2014-02-01',
               ...
               '2020-08-30', '2020-09-05', '2020-09-06', '2020-09-07',
               '2020-09-12', '2020-09-13', '2020-09-19', '2020-09-20',
               '2020-09-26', '2020-09-27'],
              dtype='datetime64[ns]', length=775, freq=None)

Fill Missing Data with ffill()

df_ffill = df.reindex(pd.date_range(start=df.index[0], end=df.index[-1])).ffill()

# Check missing values filled '2014-01-04', '2014-01-05'
display(df_ffill.loc['2014-01-02':'2014-01-06', :])
df_ffill.loc['2014-01-02':'2014-01-06', :].plot(figsize=(16, 10));

	DEXCHUS
2014-01-02	6.0504
2014-01-03	6.0505
2014-01-04	6.0505
2014-01-05	6.0505
2014-01-06	6.0524

Fill Missing Data with bfill()

df_bfill = df.reindex(pd.date_range(start=df.index[0], end=df.index[-1])).bfill()

# Check missing values filled '2014-01-04', '2014-01-05'
display(df_bfill.loc['2014-01-02':'2014-01-06', :])
df_bfill.loc['2014-01-02':'2014-01-06', :].plot(figsize=(16, 10));

	DEXCHUS
2014-01-02	6.0504
2014-01-03	6.0505
2014-01-04	6.0524
2014-01-05	6.0524
2014-01-06	6.0524

Fill Missing Data with Interpolation

Pandas Documentation

df_interpolate = df.resample('D').interpolate(method='time')

# Check missing values filled '2014-01-04', '2014-01-05'
display(df_interpolate.loc['2014-01-02':'2014-01-06', :])
df_interpolate.loc['2014-01-02':'2014-01-06', :].plot(figsize=(16, 10));

	DEXCHUS
ds
2014-01-02	6.050400
2014-01-03	6.050500
2014-01-04	6.051133
2014-01-05	6.051767
2014-01-06	6.052400