BOUNDARY AUGMENTED INPUT DECISION TREE FOR TIME SERIES DATA IMPUTATION OF IOT MICROALGAE CULTIVATION
Penulis/Author
Harnan Malik Abdullah (1); Prof. Dr. Ir. Jazi Eko Istiyanto, M.Sc. (2); Dr. techn. Aufaclav Zatu Kusuma Frisky S.Si., M.Sc. (3); Dr. Eko Agus Suyono, S.Si., M.App.Sc. (4)
Tanggal/Date
2024
Kata Kunci/Keyword
Abstrak/Abstract
Microalgae cultivation requires careful monitoring of various factors that infuence its growth. The Internet of Things (IoT) enables the automated and remote execution of these observations. Nevertheless, the Internet of Things (IoT) system also poses several issues in the context of eld operations, one of which is the potential occurrence of data shortages. Possible causes for the occurrence could include a malfunctioning sensor, issues with connectivity, and power disruptions, among other factors. The present study introduces an innovative imputation technique for the missing data utilizing a machine
learning model called Boundary Augmented Input Decision Tree (BAIDT). It is inspired by the linear imputation method that uses boundary data as input to interpolate missingdata inside the dataset. In addition, other attributes are employed as input to interpolate for the missing value. The simulation used the public ATP3 dataset of microalgae cultivation monitoring to handle the emulation of temperature missing values with various ranges. Furthermore, the proposed method was compared with a basic imputation technique named linear imputation. Based on the conducted test, the proposed technique outperformed the linear imputation even for a larger space of missing data.
