Improving AWS EC2 Spot Instance Price Prediction Accuracy using XGBoost

Dutta, Ankit (2022) Improving AWS EC2 Spot Instance Price Prediction Accuracy using XGBoost. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration manual) Download (788kB) | Preview

Abstract

Amazon Web Services offers one of the cheapest types of computing instances as spot instances which can be upto 90% cheaper that on-demand instances. This is especially beneficial to resource constrained projects. This allows users to test and develop using spot instances and finally deploying the service on on-demand or reserved instances. This allows for huge cost savings, thus making AWS spot instance price prediction with reduced error a necessity. In this evaluation 5 AWS regions were considered; namely, ‘ap-south-1’, ‘eu-west-1’, ‘us-east-1’, ‘us-west-1’, and ‘us-west-2’. The metrics used for comparison were Mean Square Error (MSE), Coefficient of Determination (r2 score), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE). After comparing these metrics for both the XGBoost and Random Forest models, it was observed that out of 20 total metrics, XGBoost had either comparable or better performance in 15 of them. The highest r2 score (0.6315) was achieved for the XGBoost model in the ‘ap-south-1’ region.

Item Type:	Thesis (Masters)
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Cloud computing
Divisions:	School of Computing > Master of Science in Cloud Computing
Depositing User:	Tamara Malone
Date Deposited:	07 Dec 2022 16:17
Last Modified:	07 Dec 2022 16:17
URI:	https://norma.ncirl.ie/id/eprint/5976

Actions (login required)

View Item