NumPy and Pandas are the two foundational Python libraries for working with data. NumPy (Numerical Python) provides fast arrays and matrices and the math to operate on them. Pandas (Python Data Analysis Library) is built on top of NumPy and is designed for tabular data, the kind you would normally keep in a spreadsheet. If you can do something in Excel, sorting, filtering, formulas, combining sheets, you can almost certainly do it in Pandas, with far more power and repeatability once your data grows large.
NumPy Arrays
A NumPy array is a grid of values of the same type that supports math on the whole array at once, which makes it much faster than a plain Python list. After importing the library with import numpy as np, you can create arrays in several ways: convert a list with np.array([1, 2, 3]), build arrays of zeros or ones with np.zeros((2, 3)) and np.ones((3, 2)), generate a range with np.arange(0, 10, 2), or produce evenly spaced values with np.linspace(0, 1, 5).
Once you have an array, you can index and slice it much like a list (my_array[0], my_array[-1], my_array[1:4]), and reshape it with my_array.reshape((2, 3)). The real advantage is element-wise math: adding, subtracting, multiplying, or dividing two arrays applies the operation to every element at once. Broadcasting extends this to arrays of different but compatible shapes, so adding a scalar to an array adds it to each element. NumPy also offers a wide set of math functions like np.exp(), np.log(), and np.sqrt() that work across an entire array.
Pandas Series and DataFrames
Pandas gives you two core structures. A Series is a one-dimensional, labeled array, created with pd.Series() from a list, tuple, or dictionary. A DataFrame is a two-dimensional, table-like structure with labeled rows and columns, the workhorse of data analysis. You can build a DataFrame from a dictionary of columns, from a NumPy array (supplying columns=[...]), or from a set of Series:
import pandas as pd
df = pd.DataFrame({
'Name': ['Alice', 'Bob', 'Charlie'],
'Age': [25, 30, 35],
'City': ['New York', 'San Francisco', 'Los Angeles']
})
Selecting, Filtering, and Modifying
Working with a DataFrame mostly comes down to getting at the right cells. Select a single column with df['Name'] and multiple columns with a list, df[['Name', 'Age']]. Filter rows with boolean conditions, such as df[df['Age'] > 25], and combine conditions with & (and) and | (or): df[(df['Age'] > 25) & (df['Age'] < 35)]. To change values, use df.at[row_label, 'Column'] for label-based access or df.iat[row_index, col_index] for position-based access. These are the everyday equivalents of clicking, sorting, and filtering in a spreadsheet.
Combining DataFrames: Concat, Merge, and Join
Real analysis usually means bringing several datasets together, and Pandas offers three main tools for it. concat stacks DataFrames, either on top of one another to add rows or side by side to add columns, which is handy when you have data of the same shape split across files. merge combines DataFrames based on shared key columns, the way a spreadsheet VLOOKUP matches rows by a common ID; you choose the join type (inner, outer, left, or right) to control which rows are kept when keys do not line up. join is a convenience method that combines DataFrames on their index rather than on a column. Together they let you assemble a single, analysis-ready table out of many smaller pieces.
Putting It Together
The pattern for almost any data task is the same: use NumPy when you need fast numerical computation on arrays and matrices, and use Pandas when you are working with labeled, tabular data. Create or load your data, select and filter to the parts you care about, apply functions and aggregations to compute what you need, and combine sources with concat, merge, or join. Master these basics and you have the toolkit to handle in Python anything you used to do in a spreadsheet, only faster, at far larger scale, and in a form you can run again whenever the data changes.