To start off, we collected data using IoT sensors from different locations of Dhaka City. The data was collected for 2 months and were stored in AdaFruit server.
Later, the data was retrieved, studied, visualised and then different Machine Learning Algorithms like SVM, Random Forest, Logistic Regression, Decision Tree, and KNN were applied to detect the anomalous data from the data received.
Detecting data anomaly in server is very important and it could come in different forms. They could arise due to failure of machines, injection of malignant data into the server, etc.
Therefore, our research is targeted to the firms in weather industry who are relying on data from different IoT sensors. The study shows that we received near to 99 percent of accuracy in detecting anomalous values.
The Notebooks can be accessed, but the paper is not made public as it has been presented to AICEC-2020.