A novel algorithm for imputing the missing values in incomplete datasets

In today’s world, our heavy reliance on digital devices for data collection has become the norm. However, when these devices fail, there is a significant risk of losing valuable information, making data mining an arduous task for data analysts. The presence of a substantial amount of missing data wi...

Full description

Saved in:
Bibliographic Details
Published in:Iran Journal of Computer Science (Online) Vol. 7; no. 1; pp. 13 - 23
Main Authors: Bhagat, Hutashan Vishal, Singh, Manminder
Format: Journal Article
Language:English
Published: Cham Springer International Publishing 01.03.2024
Springer Nature B.V
Subjects:
Algorithms
Artificial Intelligence
Computer Science
Computer Systems Organization and Communication Networks
Data analysis
Data collection
Data mining
Data Structures and Information Theory
Datasets
Mathematics of Computing
Missing data
Real time
Root-mean-square errors
Software Engineering/Programming and Operating Systems
Theory of Computation
Incomplete datasets
Data missingness
Missing values
Missingness mechanisms
Imputation
Imputing values
Data imputation model
Root mean square error
ISSN:2520-8438, 2520-8446
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In today’s world, our heavy reliance on digital devices for data collection has become the norm. However, when these devices fail, there is a significant risk of losing valuable information, making data mining an arduous task for data analysts. The presence of a substantial amount of missing data within a dataset leads to inappropriate results and incomplete data analysis. Consequently, there is a pressing need to develop an algorithm capable of efficiently and accurately predicting missing values. This research paper introduces a new algorithm, known as the IMV-RE (imputing the missing values in real-time environment) algorithm, which is based on a novel splitting approach. The IMV-RE algorithm addresses the challenge of imputing various missing values within a dataset. To enhance prediction accuracy, the algorithm sets an upper limit for each class that contains missing values, providing valuable assistance in accurately predicting these values. To evaluate the efficacy of the IMV-RE algorithm, we conducted experiments using ten benchmark datasets comprising both numerical and mixed data. The results of our comparative analysis demonstrate that the proposed IMV-RE algorithm outperforms existing techniques in terms of sensitivity to accuracy, root mean square error (RMSE), and coefficient of determination ( R 2 ).
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:2520-8438
2520-8446
DOI:10.1007/s42044-023-00154-9