|Date:||2019, December 17|
|Title:||Outlier Detection from Open Energy Data|
Open energy data consists of several features like temperature, air pressure, perpendicular sunlight, inclined sunlight, etc. collected from different geographical positions of sensors of windmills in Germany. There exists a database where all data of several years are stored. The main goal of this research is to ensure a model for that any new data which will be inserted in DB can be checked and detected if that is an outlier. Here outlier means an error of sensor or natural calamity caused data. If an outlier is inserted in the database, other applications/models of the future which will be built based on this database will lose perfection. The outliers should be marked or deleted. Four machine learning approaches will be discussed and a comparison will be done to finalize one method to implement.