<- Back to the Gallery

Alerting and Monitoring Machine Learning Model



Client Ideabytes
Professor(s) Abdullah Kadri David Lindsay
Program Computer Engineering Technology – Computing Science
Students Henry Forget, Nico Loreto, Inessa Khodak, Hadiyah Khan, Kiet Tran

Project Description:

Our project offers Ideabytes’ clients a system that can predict future values and detect performance anomalies with a decent level of accuracy using a machine learning model trained on the client’s specific IoT device. Using this machine learning model one can find previously unseen patterns and trends within datasets and make predictions of potential issues in the future, such as efficiency degradation or potentially broken equipment. The end goal is to give users a warning beforehand to avoid potentially spoiling merchandise or causing interruptions in production efficiency.

Key results of the model include:
– Learns from previously recorded data without any human guidance
– Predicts future data points with an accuracy of above 70%
– Detects anomalous data points that do not fit within the anticipated pattern
– Alerts the user if data has gone outside acceptable thresholds

A mock environment was developed to imitate the production environment of Ideabytes for the model to execute within. This way, once completed, it’d be easily integrated into Ideabytes’ existing ecosystem and immediately available for clients without any headache.


Short Description:



Contact the Team

Video Presentation


Gallery

Lists the problem statement where we must create a middleman software that interprets Ideabytes' client's data and converts it into meaningful conclusions. The model must be able to learn from previous data, predict future data trends and alert the user when data exceeds user set boundaries. Lists the use case diagram which demonstrates the responsibilities the manager (client) and the operator (Ideabytes) have. Everything required by the client can be done with installation which allows the client to receive their results without any future adjustments.
Lists the components of our mock environment. The database which stores the example data given to us by Ideabytes recorded all from one device. The algorithm takes the information, cleans it, and does calculations to convert the data to useable info. Finally the GUI presents the information to the u Shows the sequence diagram and the different API calls between each section of the system.


Funded By




Comments

Comments are closed.