Processing time with Pandas DataFrame

Pandas was created with regards to financial modeling, so as you may expect, it contains a genuinely ample number of tools for working with dates and times. Sometimes the given format of the date and time in our dataset cannot be directly used for analysis, so we pre-process these time values to obtain features like date, month, year, hours, minutes and seconds. Let’s discuss all the different ways to process date and time with Pandas dataframe. Divide date and time into multiple features: Create five dates and time using pd.date_range which generate sequences of fixed-frequency dates and time spans. Then we use pandas.Series.dt to extract the features. 

Python3

# Load library
import pandas as pd

# calling DataFrame constructor
df = pd.DataFrame()

# Create 6 dates
df['time'] = pd.date_range('2/5/2019', periods = 6, freq ='2H')
print(df['time'])  # print dataframe

# Extract features - year, month, day, hour, and minute
df['year'] = df['time'].dt.year
df['month'] = df['time'].dt.month
df['day'] = df['time'].dt.day
df['hour'] = df['time'].dt.hour
df['minute'] = df['time'].dt.minute

# Show six rows
df.head(6)

Output:

0   2019-02-05 00:00:00
1   2019-02-05 02:00:00
2   2019-02-05 04:00:00
3   2019-02-05 06:00:00
4   2019-02-05 08:00:00
5   2019-02-05 10:00:00
Name: time, dtype: datetime64[ns]


                time  year  month  day  hour  minute
0 2019-02-05 00:00:00  2019      2    5     0       0
1 2019-02-05 02:00:00  2019      2    5     2       0
2 2019-02-05 04:00:00  2019      2    5     4       0
3 2019-02-05 06:00:00  2019      2    5     6       0
4 2019-02-05 08:00:00  2019      2    5     8       0
5 2019-02-05 10:00:00  2019      2    5    10       0

Convert strings to Timestamps: We convert the given strings to datetime format using pd.to_datetime and then we can extract different features from the datetime using first method. 

Python3

# Load libraries
import numpy as np
import pandas as pd

# Create time Strings
dt_strings = np.array(['04-03-2019 12:35 PM',
                       '22-06-2017 11:01 AM',
                       '05-09-2009 07:09 PM'])

# Convert to datetime format
timestamps = [pd.to_datetime(date, format ="%d-%m-%Y%I:%M %p",
                     errors ="coerce") for date in dt_strings]

# Convert to datetimes
timestamps = [pd.to_datetime(date, format ="%d-%m-%Y %I:%M %p",
                      errors ="coerce") for date in dt_strings]

Output:

[Timestamp(‘2019-03-04 12:35:00’), Timestamp(‘2017-06-22 11:01:00’), Timestamp(‘2009-09-05 19:09:00’)]

Change the pattern of Timestamps: We can change the pattern of Timestamp using strftime method.

Python3

# Load library
import pandas as pd

# calling DataFrame constructor
df = pd.DataFrame()

# Create 6 dates
df['time'] = pd.date_range('2/5/2019', periods = 6, freq ='2H')
print("Old Pattern")
print(df['time'])  #

print("\nNew Pattern")
df['new_time']=df['time'].dt.strftime("%d-%B")
print(df['new_time'])

Output:

Old Pattern
0   2019-02-05 00:00:00
1   2019-02-05 02:00:00
2   2019-02-05 04:00:00
3   2019-02-05 06:00:00
4   2019-02-05 08:00:00
5   2019-02-05 10:00:00
Name: time, dtype: datetime64[ns]

New Pattern
0    05-February
1    05-February
2    05-February
3    05-February
4    05-February
5    05-February
Name: new_time, dtype: object

 Note: strftime method converts the timestamp to object datatype.

Extract Days Of the Week from the given Date: We use Series.dt.weekday_name to find name of the day in a week from the given Date. 

Python3

# Load library
import pandas as pd
df = pd.DataFrame()

# Create 6 dates
dates = pd.pd.Series(date_range('2/5/2019', periods = 6, freq ='M'))

print(dates)

# Extract days of week and then print
print(dates.dt.weekday_name)

Output:

0   2019-02-28
1   2019-03-31
2   2019-04-30
3   2019-05-31
4   2019-06-30
5   2019-07-31
dtype: datetime64[ns]
0     Thursday
1       Sunday
2      Tuesday
3       Friday
4       Sunday
5    Wednesday
dtype: object

Extract Data in Date and Time Ranges: We can obtain the rows that lie in particular time range from the given dataset. Method #1: If the dataset is not indexed with time. 

Python3

# Load library
import pandas as pd

# Create data frame
df = pd.DataFrame()

# Create datetimes
df['date'] = pd.date_range('1/1/2012', periods = 1000, freq ='H')

print(df.head(5))

# Select observations between two datetimes
x = df[(df['date'] > '2012-1-1 01:00:00') &
       (df['date'] <= '2012-1-1 11:00:00')]

print(x)

Output:

                 date
0 2012-01-01 00:00:00
1 2012-01-01 01:00:00                // 5 rows of Timestamps out of 1000
2 2012-01-01 02:00:00
3 2012-01-01 03:00:00
4 2012-01-01 04:00:00


                 date
2  2012-01-01 02:00:00
3  2012-01-01 03:00:00
4  2012-01-01 04:00:00
5  2012-01-01 05:00:00               //Timestamps in the given range
6  2012-01-01 06:00:00
7  2012-01-01 07:00:00
8  2012-01-01 08:00:00
9  2012-01-01 09:00:00
10 2012-01-01 10:00:00
11 2012-01-01 11:00:00

Method #2: If the dataset is indexed with time 

Python3

# Load library
import pandas as pd

# Create data frame
df = pd.DataFrame()

# Create datetimes
df['date'] = pd.date_range('1/1/2012', periods = 1000, freq ='H')

# Set index
df = df.set_index(df['date'])

print(df.head(5))

# Select observations between two datetimes
x = df.loc['2012-1-1 04:00:00':'2012-1-1 12:00:00']

print(x)

Output:

                                   date
date                                   
2012-01-01 00:00:00 2012-01-01 00:00:00
2012-01-01 01:00:00 2012-01-01 01:00:00
2012-01-01 02:00:00 2012-01-01 02:00:00
2012-01-01 03:00:00 2012-01-01 03:00:00                // 5 rows of Timestamps out of 1000
2012-01-01 04:00:00 2012-01-01 04:00:00
                                   date
date                                   
2012-01-01 04:00:00 2012-01-01 04:00:00
2012-01-01 05:00:00 2012-01-01 05:00:00
2012-01-01 06:00:00 2012-01-01 06:00:00
2012-01-01 07:00:00 2012-01-01 07:00:00
2012-01-01 08:00:00 2012-01-01 08:00:00
2012-01-01 09:00:00 2012-01-01 09:00:00               //Timestamps in the given range
2012-01-01 10:00:00 2012-01-01 10:00:00
2012-01-01 11:00:00 2012-01-01 11:00:00
2012-01-01 12:00:00 2012-01-01 12:00:00