Skip to main content

Table 1 Top 10 features selected from different models based on variable importance

From: An ensemble-based feature selection framework to select risk factors of childhood obesity for policy decision making

 

Filter (MI)

SVM-RFE

Ridge

Lasso

RandomForest

1

Age

MoDietEducation

Age

Age

SystolicPressure

2

Sleep_Normal (–)

MoRDType_LowSalt

Sex (–)

Sex (–)

MoDiastolicPressure (–)

3

BFType_Maternal (–)

RDType_2000 cal

Tobacco_No (–)

Tobacco_No (–)

MoSystolicPressure (–)

4

DiastolicPressure (–)

AdeDKnowledge

DietEducation

DietEducation

Sex

5

MoSystolicPressure

MoPE_Inadequate (–)

MoTobacco_Yes

MoDietEducation

Birthyear (–)

6

MoNumberCigarettes

DietCompliesAdvice

BFType_Maternal (–)

BFType_Maternal (–)

Tobacco_No (–)

7

Birthheight (–)

MoRDType_Free (–)

PE_Inadequate

Birthyear (–)

MoExerciseAdvice (–)

8

MoBMI

MoPEHour

MoDiabetes_No (–)

MoNumberCigarettes

MoAlcohol_No (–)

9

Birthweight (–)

DiastolicPressure (–)

PE_Adequate(–)

PE_Inadequate

PE_Inadequate

10

MoDiastolicPressure (–)

SystolicPressure

MoDietEducation

DCExecution _No

MoTobacco_Ex

  1. All “Mothers-” in the variables were replaced with “Mo-” for shorter names
  2. RDType, RecommendedDietType; MoRDType, MoRecommendedDitetType; BFType, BreastfeedingType; PE, PhysicalExercise; MoPE, MoPhysicalExercise; MoPEHour, MoPhysicalExerciseHour; AdeDKnowledge, AdequateDietaryKnowledge; DCExecution, DietCorrectExecution,