Time Series Data

What is Time Series?

Time series is data collected over time:

Daily stock prices
Monthly sales
Hourly temperature
Yearly population

The key: data is ordered by time.

Set Date as Index

For time series, make the date column your index:

code.py

import pandas as pd

df = pd.DataFrame({
    'Date': ['2024-01-01', '2024-01-02', '2024-01-03'],
    'Sales': [100, 150, 120]
})

# Convert to datetime and set as index
df['Date'] = pd.to_datetime(df['Date'])
df = df.set_index('Date')

print(df)

Output:

            Sales
Date
2024-01-01    100
2024-01-02    150
2024-01-03    120

Select by Date

code.py

# Get specific date
print(df.loc['2024-01-02'])

# Get date range
print(df.loc['2024-01-01':'2024-01-02'])

# Get specific month
print(df.loc['2024-01'])  # All of January

Resample - Change Time Frequency

Convert daily data to weekly or monthly:

code.py

# Daily sales data
df = pd.DataFrame({
    'Date': pd.date_range('2024-01-01', periods=30),
    'Sales': range(100, 130)
})
df = df.set_index('Date')

# Convert to weekly sum
weekly = df.resample('W').sum()
print(weekly)

# Convert to monthly average
monthly = df.resample('M').mean()
print(monthly)

Common Resample Frequencies

Code	Meaning
D	Day
W	Week
M	Month end
MS	Month start
Q	Quarter
Y	Year
H	Hour

Rolling Average (Moving Average)

Smooths out daily ups and downs:

code.py

df = pd.DataFrame({
    'Date': pd.date_range('2024-01-01', periods=10),
    'Sales': [100, 120, 90, 150, 130, 140, 160, 110, 180, 170]
})
df = df.set_index('Date')

# 3-day rolling average
df['Rolling_Avg'] = df['Sales'].rolling(3).mean()

print(df)

Shift Data (Lag/Lead)

Compare to previous period:

code.py

# Yesterday's sales
df['Yesterday'] = df['Sales'].shift(1)

# Tomorrow's sales
df['Tomorrow'] = df['Sales'].shift(-1)

# Change from yesterday
df['Change'] = df['Sales'] - df['Sales'].shift(1)

Percent Change

code.py

# Daily percent change
df['Pct_Change'] = df['Sales'].pct_change() * 100

print(df)

Create Date Range

code.py

# Create 7 days
dates = pd.date_range('2024-01-01', periods=7)

# Create January
jan_dates = pd.date_range('2024-01-01', '2024-01-31')

# Create weekly dates
weekly_dates = pd.date_range('2024-01-01', periods=12, freq='W')

# Create monthly dates
monthly_dates = pd.date_range('2024-01-01', periods=12, freq='MS')

Key Points

Time series has date as index
resample() changes frequency (daily→weekly)
rolling() calculates moving averages
shift() gets previous/next values
pct_change() calculates growth rate
date_range() creates date sequences

What's Next?

Learn to find and remove duplicate rows in your data.