| Efficient smartification in buildings pre-requires highly reliable data. Modern AI applications consider real-time emergent responsiveness, which renders control safety of outmost importance. A number of sensors and other devices are installed that retrieve data on energy consumption, indoor conditions, and other information. By analyzing this data, smart home devices are able to customise user experiences while optimizing energy use while enhancing security in a real-time manner. In this paper, a holistic data treatment framework for Smart Building IoT applications is presented. The proposed framework considers a multi-factorial anomaly detection and treatment. The implemented functional filtering pipeline consists of statistical filtering rules for detecting, recognizing and mitigating the most common anomalies observed in the historical data. First a methodology for estimating missing data values based on available data is considered. Moreover, an outlier detection and healing mechanism is also integrated to improve the accuracy and reliability of the analysed results.The results showcase that the clean data are of better quality for further exploitation. |
*** Title, author list and abstract as seen in the Camera-Ready version of the paper that was provided to Conference Committee. Small changes that may have occurred during processing by Springer may not appear in this window.