Data Analysis using Numpy

1. Introduction

NumPy provides efficient multidimensional arrays for numerical work. Unlike a Python list, an array typically stores values of a common data type and supports operations across many values without writing a Python loop for each element.

2. Core concepts

Shape describes the length of each dimension; dtype describes the element representation. Indexing selects values, slicing selects ranges, and an axis tells a reduction which dimension to collapse. Broadcasting lets compatible shapes participate in elementwise operations without explicitly copying a repeated value.

3. Topics covered in this module

• Array creation, shape, dimensions, dtype, and conversion
• Indexing, slicing, masks, multidimensional axes, and views
• Broadcasting rules and vectorized arithmetic
• Aggregations, missing values, numeric precision, and built-in functions

4. Syntax and how to read it

import numpy as np

scores = np.array([[70, 80], [90, 100]], dtype=float)
column_means = scores.mean(axis=0)
print(column_means)

axis=0 reduces rows and leaves one mean per column. The axis choice follows the data shape, not a universal rule.

5. Practical workflow

Inspect shape and dtype, verify that dimensions represent the intended observations and features, then apply a vectorized operation. Test a small array by hand and watch for integer overflow, NaN propagation, and unintended view mutation.

6. Common mistakes

• Using the wrong axis and obtaining a plausible but incorrectly grouped result.
• Assuming a slice is an independent copy when NumPy may return a view.

7. Best practices

• Check shape, dtype, and units before arithmetic.
• Use explicit axis arguments and small hand-calculated tests.

8. Practice task

Given a 4-row by 3-column array, which axis produces one mean for each of the three columns?

9. One possible solution

Use axis=0; it reduces across rows and leaves the three column means.

10. Summary

Use typed arrays, vectorized operations, indexing, axes, broadcasting, and numerical functions. Check shape, dtype, and units before arithmetic. Use explicit axis arguments and small hand-calculated tests.

11. What comes next

The next module, Data Analysis using Pandas, builds on this foundation and applies it to a new part of the analytics workflow.