High cardinality categoricals

32. High cardinality categoricals

easy

General

senior

Handle high-cardinality categorical features during preprocessing, where categories can have many unique values and naive one-hot encoding explodes feature size. You’ll implement a simple top-K + “OTHER” encoder to keep the representation compact.

The encoding is defined as:

\text{enc}(x)= \begin{cases} x & \text{if } x \in \text{TopK}(X, K) \\ \text{"OTHER"} & \text{otherwise} \end{cases}

Requirements

Implement the function:

python

Rules:

Count category frequencies in values and identify the k most frequent categories.
If there are ties in frequency, break ties by lexicographic order (ascending).
Replace any category not in the top-k set with the string "OTHER".
Return the encoded array in the same order as input.
Use only NumPy.

Example

python

Output:

python

Input Signature

Argument	Type
k	int
values	np.ndarray

Output Signature

Return Name	Type
value	np.ndarray

Constraints

Use only NumPy.
Return np.ndarray in same input order.
Tie-break: lexicographic ascending on equal counts.

Hint 1

Use np.unique(values, return_counts=True) to get unique categories and their frequencies.

Hint 2

Use np.lexsort to sort indices based on multiple keys. Remember it sorts by the last key passed as primary. To sort by frequency descending and then value ascending, pass keys corresponding to value and negative frequency.

Hint 3

Use np.isin to create a boolean mask for the original array indicating which elements are in the top-k set, then use np.where to replace non-top-k elements with "OTHER".

Roles

ML Engineer

AI Engineer

Data Scientist

Quantitative Analyst

Companies

General

Levels

senior

entry

Input Arguments

Edit values below to test with custom inputs

You need tolog in/sign upto run or submit