Releases: JuliaAI/ScientificTypes.jl
Releases · JuliaAI/ScientificTypes.jl
v0.8.1
v0.8.0
v0.7.2
v0.7.1
v0.7.0
v0.6.1
v0.6.0
v0.6.0 (2020-01-29)
Merged pull requests:
v0.5.1
- Remove
export MLJ
to avoid conflicts with the package of the same name, which uses ScientificTypes
v0.5.0
- (Breaking) To address persistent performance issues, the definition of
scitype
for arrays is changed. The new behaviour only effects arrays whoseeltype
is aUnion{Missing, _ }
type. The new complete definition is: - The scitype of an
AbstractArray
,A
, is alwaysAbstractArray{U}
whereU
is the union of the scitpyes of the elements ofA
, with one exception: Iftypeof(A) <: AbstractArray{Union{Missing,T}}
for someT
different fromAny
, then the scitype ofA
isAbstractArray{Union{Missing, U}}
, whereU
is the union over all non-missing elements, even ifA
has no missing elements.
To force the old behaviour (no exception) use scitype(A, tight=true)
.
- (Breaking) To improve performance, the behaviour of
coerce(A , S)
is modified in the case of arrays whoseeltype
is aUnion{Missing, _ }
type. In these cases the return value haselscitype
equal toUnion{Missing, S}
, even ifA
has no missing values. To force the old behaviour, usecoerce(A, S, tight=true)
. Query?coerce
for details.