Influence of Feature Selection Methods on Cardiotocography Data: A Quantitative Investigation
M. Ramla
M. Ramla, Department of Computer Applications, Faculty of Science and Humanities, SRM IST, Kattankulathur, Chennai (Tamil Nadu), India.
Manuscript received on 18 July 2019 | Revised Manuscript received on 25 July 2019 | Manuscript Published on 01 August 2019 | PP: 19-23 | Volume-8 Issue-4S2, April 2019 | Retrieval Number: D10060484S219/19©BEIESP | DOI: 10.35940/ijeat.D1006.0484S219
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (

Abstract: Cardiotocography is the most powerful tool to monitor the fetal health state during both antenatal and postnatal periods. Using the EFM machine, the fetal heart beat pace and uterine contractions of pregnant mother are recorded simultaneously. This signal data will help doctors to assess wellness of the fetus and classify them. Feature selection is of utmost importance to potentially uplift the predictive model for CTG analysis. The objective of Feature Selection is to keep more pertinent features that capture the hidden insights and ignoring the redundant variables to improve the classification accuracy of the predictive model. A walkthrough, on the UCI-CTG dataset using different feature selection methods such as filters, wrappers, and embeds are made and then compared. This paper attempts to analyze the inherent nature of the data using methods such as Linear Discriminant Analysis, Recursive Elimination, Forward and Backward Elimination and Lasso Regression. The extracted features are used for classification and the performance evaluation of the classifier is observed using Accuracy as the performance metrics.
Keywords: Feature Selection, CTG, Fetal Heart Rate, LDA, Lasso, Recursive Feature Elimination.
Scope of the Article: Data Analytics