Machine learning to predict unintended pregnancy among reproductive-age women in Ethiopia: evidence from EDHS 2016

Table 4 Compares selected machine learning models in choosing the top features

R.No	Top Features	ETC	GB	RF	XGB	DT	LR	Median
1	Region	0.0727	0.1047	0.0212	0.0336	0.1181	0.0308	0.0727
2	Ideal number Children	0.0601	0.0613	0.0495	0.1558	0.0557	0.5931	0.0601
3	Wealth index	0.0459	0.0461	0.0318	0.0472	0.0469	0.3090	0.0461
4	Husband education	0.0423	0.0461	0.0220	0.0107	0.0478	0.0279	0.0423
5	Religion	0.0468	0.0476	0.0218	0.0162	0.0418	0.1262	0.0418
6	Age at first sex	0.0417	0.0391	0.0271	0.0488	0.0411	0.3169	0.0411
7	Total birth	0.0378	0.0363	0.0246	0.0155	0.0418	0.0497	0.0364
8	Refuse sex	0.0379	0.0364	0.0194	0.0010	0.0344	0.0756	0.0344
9	Age at 1st birth	0.0344	0.0329	0.0330	0.0666	0.0362	0.3659	0.0344
10	Maternal Educational status	0.0335	0.0337	0.0231	0.0149	0.0373	0.1667	0.0336

RF: Random Forest; ExtraTrees; DT: Decision Tree; LR: Logistic Regression; GB: Gradient Boost; XGBoost: Extreme Gradient Boosting

ISSN: 1472-6874