Python datetime Review

The datetime library in Python helps manage and manipulate date and time. Here’s a quick review:

Creating a datetime object

from datetime import datetime
 
# Specify year, month, day, and optionally hour, minute, and second
my_year = 2017
my_month = 1
my_day = 2
my_hour = 13
my_minute = 30
my_second = 15
 
# Creating a basic date (defaults to 00:00 time)
my_date = datetime(my_year, my_month, my_day)
print(my_date)
 
# Creating a full datetime with time
my_date_time = datetime(my_year, my_month, my_day, my_hour, my_minute, my_second)
print(my_date_time)

Accessing parts of a datetime object

You can retrieve specific components like the day or hour:

print(my_date.day)       # 2
print(my_date_time.hour) # 13

---

Pandas and Datetime

Pandas makes it easy to work with datetime objects, especially when dealing with time-series data.

Converting Strings to Datetime

Often, dates in datasets are stored as strings. Use pd.to_datetime to convert them to datetime objects.

import pandas as pd
 
# Example Series with dates
myser = pd.Series(['Nov 3, 2000', '2000-01-01', None])
print(myser)
 
# Convert to datetime
converted = pd.to_datetime(myser)
print(converted)
 
# Handling ambiguous dates (e.g., European format)
euro_date = '10-12-2000'
print(pd.to_datetime(euro_date, dayfirst=True))  # Interprets as 10th December 2000

Custom Time String Formatting

If the date format is non-standard, you can specify it explicitly using the format parameter. This can improve parsing performance.

style_date = '12--Dec--2000'
parsed_date = pd.to_datetime(style_date, format='%d--%b--%Y')
print(parsed_date)

Reference for format codes: Python Datetime Format Codes

---

Working with Time-Series Data

Example Dataset: Retail Sales

Let’s load a dataset:

sales = pd.read_csv('RetailSales_BeerWineLiquor.csv')
print(sales.head())
 
# Convert DATE column to datetime
sales['DATE'] = pd.to_datetime(sales['DATE'])
print(sales.dtypes)

Setting Datetime Index

For time-series operations, set the datetime column as the index:

sales = sales.set_index("DATE")
print(sales.head())

Attempt to Parse Dates Automatically

The parse_dates parameter is a powerful feature in Pandas’ data reading functions (like pd.read_csv or pd.read_excel) that allows you to automatically convert columns containing date strings into datetime objects when loading the data. This saves you the extra step of manually calling pd.to_datetime() on those columns after loading the file.

sales = pd.read_csv('RetailSales_BeerWineLiquor.csv',parse_dates=[0])

---

Resampling Time-Series Data

The resample() method is used for aggregating data based on time frequency.

Resampling Rules (Offset Aliases)

Common aliases for resampling:

• A: Year-end frequency
• Q: Quarter-end frequency
• M: Month-end frequency
• W: Weekly frequency
• D: Daily frequency

Full list: Pandas Offset Aliases

Example: Yearly Means

# Resample by year and calculate the mean
yearly_mean = sales.resample('A').mean()
print(yearly_mean)

Using the .dt Accessor

Once a column is in datetime format, you can access various attributes and methods:

sales = sales.reset_index()
 
# Extracting components
print(sales['DATE'].dt.month)        # Extract month
print(sales['DATE'].dt.is_leap_year) # Check for leap year