ts_intro.py

# we have two time series of power and oil production by month in Russia in 2014, TWh (млрд. кВтч) and mln t

""" Electricty production by month, TWh 
2014-01-31    102.0
2014-02-28     93.5
2014-03-31     94.3
2014-04-30     84.4
2014-05-31     79.6
2014-06-30     74.6
2014-07-31     77.5
2014-08-31     78.8
2014-09-30     80.1
2014-10-31     92.8
2014-11-30     96.0
2014-12-31    104.0
"""

z = """Oil production by month, mln t 
2014-01-31    42.6
2014-02-28    38.3
2014-03-31    42.3
2014-04-30    41.0
2014-05-31    42.5
2014-06-30    41.3
2014-07-31    42.3
2014-08-31    42.4
2014-09-30    41.4
2014-10-31    42.5
2014-11-30    41.4
2014-12-31    42.5
"""

import pandas as pd

# datapoints as vector
power_prod_vector = [102.0, 93.5, 94.3, 84.4, 79.6, 74.6, 77.5, 78.8, 80.1, 92.8, 96.0, 104.0]
oil_prod_vector = [42.6, 38.3, 42.3, 41.0, 42.5, 41.3, 42.3, 42.4, 41.4, 42.5, 41.4, 42.5]

assert len(power_prod_vector) == 12
assert len(oil_prod_vector) == 12

#creating Series objects
ts_power = pd.Series(power_prod_vector)
ts_oil   = pd.Series(oil_prod_vector)

# this data above does not have time labels, which is essential for time sereis

#
# note: here you may refer to dates_intro.py to refresh date types in python
#

# we can generate time index using pandas function date_range()
# in example below it will gerentate end of month dates
# note: for small class discussion: why end-of-month is better as a time marker for monthly data?
time_index = pd.date_range('1/1/2014', periods=12, freq='M')

# recreating time series with time index
ts_power = pd.Series(power_prod_vector, time_index)
# alternatinely can assign index to exiting time series
ts_oil.index = time_index

#congratulations, ts_power and ts_oil are valid Timeseries variables now

# can plot a variable:
ts_power.plot()


# time series can be handled together as dataframe, which is a basic 
# data structure for dealing with several time series 

prod_df = pd.DataFrame({'power' : ts_power,
                          'oil' : ts_oil})

print("\n*** Dataframe sample ***")
print(prod_df)

prod_df.plot()

# todo: нарисовать отклонения от cреднегодового уровня обеих переменных + сравниь какая переменная более волатильна