Feature Selection In The Human Aspect of Information Security Questionnaires Using Multicluster Feature Selection

Abstract

The human aspect is a major vulnerability in information security that relates to
knowledge, attitude, and behavior in information security. This study aims to determine
the essential variables used to assess the level of information security in a person. This
study used Principal Component Analysis (PCA) and Multicluster Feature Selection
(MCFS) approaches. Data collection using questionnaires based on The Human Aspects
of Information Security Questionnaire (HAIS-Q) with adaptation to the Indonesian
environment. The questionnaire distributed to 148 employees at XYZ institution as
respondents. The level of information security applied to each individual based on the
result of respondents' responses to the questionnaire. Data processing has been
conducted using the Python programming language and Notebook Juypter as tools. The
aim of PCA is to help in reducing some features with a specific explained variance ratio.
MCFS is used to help reduce the number of features which can represent all features on
HAIS-Q. The selected features based on knowledge are password security, social media,
mobile device, and incident report. The selected features based on personal attitudes are
password security, email usage, internet usage, mobile device, information handling, and
incident reporting. The selected features based on behavior are password security, email
used, internet used, mobile device, information handling, and incident reporting