alteryx/evalml

Data Health: Sparsity Refactor

Aberta

#1.801 aberto em 9 de fev. de 2021

 (2 comentários) (1 reação) (1 responsável)Python (93 forks)auto 404
enhancementgood first issue

Métricas do repositório

Stars
 (852 estrelas)
Métricas de merge de PR
 (Métricas PR pendentes)

Description

Currently, the sparsity score is calculated using a discrete count of values as a threshold. It might make more sense to refactor the SparsityDataCheck to use a relative threshold instead of a fixed count threshold.

SparsityDataCheck.sparsity_score(col, relative_count_threshold=0.10):
    <implementation>

Here the new relative_count_threshold should be a percentage of the total length of the column.

Guia do colaborador