Implement np.random.choice
Quick Overview
This question evaluates understanding of random sampling algorithms, probability distributions (uniform and weighted), handling of weights and edge-case validation, and engineering trade-offs in data-structure-aware implementation for both integer ranges and explicit collections.
Pinterest
Feb 1, 2026, 6:03 PM
Data Scientist
Technical Screen
Coding & Algorithms
3
0
Loading...
Implement a function similar to numpy.random.choice.
Function signature
choice(a, size=None, replace=True, p=None) -> samples
Requirements
-
acan be:-
an integer
nmeaning sample from[0, 1, ..., n-1], or - a list/array of items to sample from.
-
an integer
-
sizecan be:-
None(return a single item), or -
an integer
k(return a list of lengthk).
-
-
replace:-
if
True, sampling is with replacement. -
if
False, sampling is without replacement (no duplicates). Assumek <= len(a).
-
if
-
p:-
optional probability weights aligned with
a. -
if
None, use uniform distribution.
-
optional probability weights aligned with
Tasks
- Describe an algorithm to sample with replacement (uniform and weighted).
- Describe an algorithm to sample without replacement (uniform and weighted).
-
Discuss time complexity and edge cases (invalid
p, floating point sum != 1,size=None, etc.).