Data Science Essentials in Python

Unit 25

Demystifying Universal Functions

Vectorized universal functions, or ufuncs, are a functional counterpart of broadcasting. You can apply ufuncs to all array items at once in one function call. numpy provides a lot of ufuncs, and here are some of them:

• arithmetic: add, multiply, negative, exp, log, sqrt

• trigonometric: sin, cos, hypot

• bitwise: bitwise_and, left_shift

• relational and logical: less, logical_not, equal

• maximum and minimum

• functions for working with floating-point numbers: isinf, isfinite, floor, isnan

Let’s say we recorded the stock prices for the eight sap symbols after and before the weekend of 01/10/2016 in a one-dimensional array called stocks:

stocks

array([ 140.49,    0.97,   40.68,   41.53,   55.7 ,   57.21,   98.2 ,