Abstract:
Abstract—Health systems bear a crucial responsibility for the
well-being of individuals across their entire lifespan. Accurate
prediction of the delivery date for childbirth is of paramount
importance for both the expectant mother and her infant. It
plays a vital role in ensuring the safety and health of both
parties. A miscalculated decision in this regard not only increases
the risks for the mother’s life but also poses potential harm
to the newborn. To address this issue and mitigate associated
risks, we have developed a prediction model based on machine
learning techniques. In this research study, we utilized the
Ethiopian Demographic and Health Survey (EDHS) dataset to
predict the delivery date for childbirth and create our model.
We follow experimental research design methodology as a result
we employed an experiment regression-based machine learning
algorithms, including Linear Regression, Multilayer Perceptron,
and Random Forest, to develop the predictive model. Once the
model was constructed, we evaluated its performance using well
established evaluation metrics, such as mean absolute error,
mean squared error, and R2 score. Upon careful examination
of the evaluation results, we determined that the R2 score served
as a reliable metric to assess the performance of our models.
The obtained R2 scores were 0.35 for Linear Regression, -25.1
for Multilayer Perceptron, and 0.67 for Random Forest. Based
on these scores, it became evident that the Random Forest
algorithm exhibited superior performance compared to the othe
