The purpose of fetal health classification is to identify any potential risks or complications that may require medical intervention. It helps healthcare providers make informed decisions regarding the timing and type of interventions, such as medical treatments, surgical procedures, or specialized care during pregnancy or after birth.
The classification of fetal health is typically based on established guidelines and criteria developed by medical professionals and organizations. These guidelines consider factors such as fetal size, accelerations, uterine contractions, organ function, and fetal movements
The project begins with data exploration and preprocessing, addressing issues. It then delves into the creation and evaluation of different machine learning models, both linear and non-linear, to determine their effectiveness in fetal health classification.
Linear classifiers, such as Logistic Regression, and Support Vector Machines (SVMs), are employed initially to establish a baseline for model performance. Non-linear classifiers, including Random Forest, and k-nearest neighbors KNN, to potentially improve classification accuracy.
Model Logistic Regression with accuracy 0.903 and with F1_SCore 0.89. Model SVM with accuracy 0.903 and with F1_SCore 0.900 Model KNN with accuracy 0.936 and with F1_SCore 0.933 Model Random Forest with accuracy 0.97 and with F1_SCore 0.971 🎇