Skip to main content

Table 5 Explanation of the local contribution of sociodemographic features through the Eli5 model in predicting the level of preschool DD in a single test instance using the LightGBM model. The presence of a question mark in a table generated by LIME may signify uncertainty or a need for further clarification regarding the results or interpretations provided

From: Improving machine learning models through explainable AI for predicting the level of dietary diversity among Ethiopian preschool children

y = 0 (probability 0.995, score -5.310) top features

Contribution?

Feature

Value

+ 1.613

Wealth Index

0.1517

+ 0.986

Region

0.1575

+ 0.565

Age of child

0.1020

+ 0.338

Age in 5-year group

0.1136

+ 0.141

Sex of child

0.0251

+ 0.099

Mother Occupation

0.0286

+ 0.090

Frequency of listening to Radio

0.0536

+ 0.073

Contraception Methods

0.0296

+ 0.061

Husband Education Level

0.0695

+ 0.019

Mother Education Level

0.0607

+ 0.007

Breastfeeding

0.0345

-0.046

Body Mass Index

0.0586

-0.118

Mother Religion

0.0643

-0.179

Source of Drinking Water

0.0333