Introduction to Pandas — Series

Pandas is a Python-based data processing and analysis tool that provides efficient and flexible data structures, making it easier to work with structured data such as tables or time series. Built on top of NumPy, Pandas leverages its efficient numerical computing capabilities while offering higher-level interfaces like DataFrame and Series, enabling more intuitive and powerful data manipulation. This article introduces the Series object in Pandas.

Series

#

A Series is a one-dimensional data structure with an index, similar to an array with named labels. Each data value has a corresponding index, which allows for quick access and manipulation. Even if the order of data changes, you can still locate values easily using their index. Creating a Series is simple:

import pandas as pd

data = pd.Series([10, 20, 30, 40, 50])
print(data)

This outputs:

0    10
1    20
2    30
3    40
4    50
dtype: int64

The left column shows the index, and the right column shows the values. You can also define your own custom index:

import pandas as pd

data = pd.Series([10, 20, 30, 40, 50], index=['a', 'c', 'b', 'f', 'k'])
print(data)

This will output:

a    10
c    20
b    30
f    40
k    50
dtype: int64

With custom indexing, you can retrieve elements using their labels (e.g., data["a"]). You can still use the original index as well. To distinguish between them:

• Positional Indexing: e.g., data[1:3] — excludes the end index
• Label Indexing: e.g., data["a":"c"] — includes the end index

You can also use a list to index:

data[["f", "a", "k"]]

Output:

f    40
a    10
k    50
dtype: int64

loc and iloc

#

You can use integers as labels too, which might lead to confusion. Here’s what happens:

• data[...] uses label-based indexing when accessing a single element
• data[...] uses positional indexing when slicing

To avoid this confusion, Pandas provides:

• loc for label-based indexing
• iloc for position-based indexing

Example:

data.iloc[3]

Conditional Filtering

#

Just like NumPy arrays, you can filter Series with conditions:

data[(data > 20) | (data < 40)]

Creating a Series from a Dictionary

#

You can also create a Series by passing a dictionary:

data = pd.Series({
    "Lettuce": 4.1,
    "Daikon": 2.2,
    "Tomato": 5.3,
    "Potato": 3.7,
    "Cucumber": 6.8
})

The keys become the index.

Arithmetic Operations

#

You can perform arithmetic on Series. Only elements with matching indices are computed together; others result in NaN:

s1 = pd.Series([1, 2, 3, 4, 5], index=['a', 'b', 'c', 'd', 'e'])
s2 = pd.Series([50, 40, 30, 20, 10], index=['a', 'b', 'd', 'e', 'f'])

s1 + s2

Results in:

a    51.0
b    42.0
c     NaN
d    34.0
e    25.0
f     NaN
dtype: float64

To fill missing values with a default value:

s1.add(s2, fill_value=0)

Results in:

a    51.0
b    42.0
c     3.0
d    34.0
e    25.0
f    10.0
dtype: float64

Common Methods

#

• min, max, mean, sum: basic statistics
• describe(): returns a summary of statistics

Applying Functions to Each Element

#

To apply a function to each element (e.g., assigning grades based on scores):

def grade_from_score(score):
  if score >= 80:
    return "A"
  elif score >= 60:
    return "B"
  else:
    return "C"

Use apply to apply the function:

grades = scores.apply(grade_from_score)

You can also pass a lambda (anonymous) function — see the blog’s related article for more.