Oversampling Response Stretch based Fetal Health Prediction using Cardiotocographic Data

Abstract

          With the current development of innovation towards medication, different ultrasound strategies are accessible to discover the fetal wellbeing. It is analyzed with different clinical parameters with 2-D imaging and other test. However, wellbeing expectation of fetal heart still remains an open issue due to unconstrained exercises of the hatchling, the minor heart estimate and insufficiency of information in fetal echocardiography. The machine learning methods can discover out the classes of fetal heart rate which can be utilized for prior estimating. With this outline, we have utilized Cardiotocographic Fetal heart rate dataset extricated from UCI Machine Learning Store for foreseeing the fetal heart rate wellbeing classes. The Prediction of fetal health rate are accomplished in six ways. Firstly, the data set is preprocessed with Feature Scaling and lost values. Secondly, exploratory data analysis is done and the distribution of target feature is visualized. Thirdly, the raw data set is fitted to all the classifiers and the performance is analysed before and after feature scaling. Fourth, the raw data set is subjected to oversampling methods like Random Oversampler, SMOTE, Borderline SMOTE, KMeansSMOTE, SVMSMOTE and ADASYN. Fifth, the oversampled dataset by above mentioned methods are fitted to all the classifiers and the performance is analyzed before and after feature scaling. Sixth, performance analysis is done using metrics like Precision, Recall, F-score, Accuracy and running time. The execution is done using python language under Spyder platform with Anaconda Navigator. Experimental results shows that the Random Forest and Decision Tree classifier tends to retain 93% and 92% accuracy respectively before and after feature scaling. The Random Oversampled dataset shows that the Random Forest and Decision Tree classifier tends to retain 99% and 98% accuracy respectively before and after feature scaling. The SMOTE, Borderline SMOTE, KMeansSMOTE, SVMSMOTE and ADASYN resampled dataset shows that the Random Forest and Decision Tree classifier tends to retain 97% and 96% accuracy respectively before and after feature scaling.