Skip to main content

Table 3 Important features based on feature selection methods

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

Methods Used

Features

Accuracy

Mutual Information

‘Age of child’, ‘Birth order’, ‘Mother education Level’, ‘Mother occupation’, ‘Wealth Index’, ‘Place of Residence’, ‘Body mass index’, ‘Type of Cooking ‘, ‘Type of Cooking’, ‘Contraception Method’, ‘husband Education Level’,' Frequency of listening Radio ‘,' Frequency of listening Television’, ‘Region ‘, ‘Frequency of Reading Newspaper’,' ‘currently breastfeeding’.

97.8%

Chi-square Test

‘Age of child’, ‘Birth order’, ‘Mothers Education Level’, ‘Mothers Occupation’, ‘Wealth Index’, ‘Place of Residence’, ‘Body mass index’, ‘Type of Cooking ‘, ‘Contraception Method’, ‘husband Education Level’,' Frequency of listening Radio ‘,' Frequency of listening Television ‘, ‘Frequency of Reading Newspaper’, Region, ‘’currently breastfeeding’

98.8%

Step Backward Feature Selection

Sex of child’, ‘age in 5-year group’, ‘Age of child’, ‘Religion’, ‘Mother education Level’, ‘Mother occupation’, ‘Wealth Index’, ‘body mass index’, ‘Contraception Method’. ‘Husband Education Level’.' ‘Frequency of listening to Radio’, ‘Region ‘, ‘currently Breastfeeding’, ‘Source of Drinking Water’, ‘Presence of diarrhea’

99.3%