Evaluating Imputation Strategies for Handling Missing Data: A Comparative Study

Missing data is a significant challenge across various domains of data analysis, impacting the accuracy of analysis and interpretation of underlying patterns and relationships within datasets. This study specifically focuses on two different datasets: addressing the absence of depth values in cattle backbone data captured using 3-D cameras and addressing missing data in real-time recordings of RR intervals in fetal health rate variability (FHRV) obtained from the sensors used for internal monitoring of electrocardiogram (ECG) recordings taken prior to fetal delivery. To tackle these gaps, popular time series imputation techniques, including linear interpolation, spline interpolation, and autoregressive models, are employed. The performance of each model is evaluated using the root mean square error (RMSE). This study ultimately selects the optimal model for handling the missing data which is important for data analysis research work.