1 -0.788818 -0.985819 0.403796 -1.753784
2 -0.155881 -1.752672 1.037444 -0.400793
3 -0.777546 1.730278 0.417114 0.184079
4 -1.763660 -0.375469 0.098678 -1.553824
5 -1.134258 1.401821 1.227124 0.979389
6 0.458838 -0.143187 1.565701 -2.085863
7 -0.103058 -0.366170 -0.478036 -0.032810
8 1.040318 -0.128799 0.786187 0.414084
The column names provide an efficient mechanism to access data in the DataFrame object,
again similar to structured arrays:
In [ 23 ]: df[‘No2’][ 3 ] # value in column No2 at index position 3
Out[23]: 1.7302783624820191
To work with financial time series data efficiently, you must be able to handle time indices
well. This can also be considered a major strength of pandas. For example, assume that
our nine data entries in the four columns correspond to month-end data, beginning in
January 2015. A DatetimeIndex object is then generated with date_range as follows:
In [ 24 ]: dates = pd.date_range(‘2015-1-1’, periods= 9 , freq=‘M’)
dates
Out[24]: <class ‘pandas.tseries.index.DatetimeIndex’>
[2015-01-31, ..., 2015-09-30]
Length: 9, Freq: M, Timezone: None
Table 6-2 lists the parameters that the date_range function takes.
Table 6-2. Parameters of date_range function
Parameter Format Description
start
string/datetime
left bound for generating dates
end
string/datetime
right bound for generating dates
periods
integer/None
number of periods (if start or end is None)
freq
string/DateOffset
frequency string, e.g., 5D for 5 days
tz
string/None
time zone name for localized index
normalize
bool, default None
normalize start and end to midnight
name
string, default None
name of resulting index
So far, we have only encountered indices composed of string and int objects. For time
series data, however, a DatetimeIndex object generated with the date_range function is
of course what is needed.
As with the columns, we assign the newly generated DatetimeIndex as the new Index
object to the DataFrame object:
