A Framework for Implementing Prediction Algorithm over Cloud Data as a Procedure for Cloud Data Mining

Safwan A S Al-Shaibani

doi:10.54060/jieee/002.02.021

Abstract

The cloud has become an important phrase in data storage for many reasons. Cloud services and applications are widespread in many industries including healthcare due to easy access. The limitless quantity of data available on the clouds has triggered the interest of many researchers in the recent past. It has forced us to deploy machine learning for analyzing the data to get insights as well as model building. In this paper, we have built a service on Heroku Cloud which is a cloud platform as a service (PaaS) and has 15 thousand records with 25 features. The data belongs to healthcare and is related to post-surgery complications. The boost prediction algorithm was applied for analysis and implementation was done in python. The results helped us to determine and tune some of the hyperparameters which have correlations with complications and the reported accuracy of training and testing was found to be 91% and 88% respectively.

Highlights

In the last decade, people have migrated to the cloud for hassle free storage of data
Data mining techniques implemented through cloud computing will allow us to retrieve useful information from virtually integrated data while lowering infrastructure and storage expenses
We take up a case of health care data to demonstrate how cloud data can be data mined? In the healthcare sector, data mining and machine learning have endless applications

Summary

Introduction

People have migrated to the cloud for hassle free storage of data. Data mining and machine learning have endless applications These can help to streamline hospital administrative procedures, map and manage infectious diseases, customize medical treatments,etc[2]. They will play a key role in supporting clinical decision-making, allowing for earlier disease identification, and tailored treatment plans to ensure optimal outcomes. These may be used to explain and advise patients with various care choices on possible disease pathways and outcomes [1,2].

Methods

Results

Conclusion