dotnet / dotnet/machinelearning

Vector of length 0 as missing value for vector type columns is problematic

Open
#1,990 2 comments 0 reactions 1 assignee Claimed by @yaeldekel View on GitHub
area-Transforms bug Priority:1
Dominant language
C#
Stars
9.4k
Forks
2k
Avg merge
2d 20h
Merged PRs (30d)
11

Description

ValueMappingTransformer returns a vector of length 0 when the value it is trying to map is not in the dictionary. When data that has a missing value like this is saved as text data, then TextLoader is not able to reload this data. For example: if the mapping is
"a" -> 0,1
"b" -> 1,0
"c" -> 1,1

and we apply the ValueMappingTransformer to the following data:

a b
b c
d a

the result should be:

a b 0 1 1 0
b c 1 0 1 1
d a 0 1

but the actual result is
a b 0 1 1 0
b c 1 0 1 1
d a 0 1

which causes TextLoader to load the data as

a b 0 1 1 0
b c 1 0 1 1
d a 0 1 0 0

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.