alteryx / alteryx/woodwork

SPIKE: Investigate inference performance improvements

未关闭
#1,133 0 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看
主要语言
Python
星标
155
派生
24
PR 合并指标
30 天内没有已合并 PR

描述

The current type inference process in Woodwork can likely be improved (runtime). Currently when performing type inference, Woodwork will execute all of the inference functions defined on the "root types", which are defined as logical types that do not have a parent logical type defined in the relationships. (Root types can be viewed through `ww.type_system.root_types`)

In some circumstances this process can be wasteful as we end up calling inference functions that we know will not return a match. For example, if a column contains integer values, and the integer inference function returns True, we know that this series cannot contain things like URLs or EmailAddresses, so executing those inference functions is unnecessary.

We should investigate approaches to improve Woodwork type inference performance by eliminating unnecessary inference function calls.

贡献指南

打开贡献指南

评估

这个 Issue 还没有评估数据。

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。