Ramadlani, Muhammad Faris Ruri (2019) Seleksi fitur menggunakan algoritma chi square untuk prediksi cacat perangkat lunak. Undergraduate thesis, Universitas Islam Negeri Maulana Malik Ibrahim.
|
Text (Fulltext)
14650052.pdf - Accepted Version Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (3MB) | Preview |
Abstract
INDONESIA:
Cacat perangkat lunak atau biasa yang dikenal dengan istilah bug atau kesalahan di dalam aplikasi yang sudah dibuat oleh seorang pengembang atau programmer. Atau sesuatu yang muncul ketika hasil yang diharapkan dari sebuah perangkat lunak bertolak belakang hasil yang sebenarnya. Namun , bisa juga saat pengujian menemukan sesuatu dalam sistem yang menyimpang dari perilaku yang diharapkan , bukan berarti bisa dikatkan sepenuhnya bahwa hal tersebut merupakan bug. Data yang digunakan adalah dataset dari NASA, yang berjumlah 5 data, yaitu CM1, JM1, KC1, KC2, PC1.
Penelitian ini melakukan pemilihan fitur menggunakan seleksi fitur dengan algoritma Chi Square. Uji coba dilakukan dengan menyeleksi fitur fitur yang ada kemudian menghitung akurasi dengan metode klasifikasi Naïve Bayes. Diambilnya berbagai jumlah yang berbeda pada setiap data, akan diketahui nantinya jumlah fitur dan fitur apa saja yang memiliki pengaruh signifikan pada penelitian ini.
Hasil dari penelitian ini menyimpulkan bahwa fitur-fitur yang memiliki pengaruh signifikan terhadap prediksi cacat perangkat lunak berdasarkan seleksi fitur Chi Square pada dataset CM1 adalah fitur i (Intelligence), t (Time to write program) dengan akurasi 93,47%. Dataset JM1 adalah fitur i (Intelligence), l (Program length), total_Opnd (Total operand), d (Difficulty), v (Volume), b (Error estimate), loc (Line of code), v(g) (Cyclomatic complexity), ev(g) (Essential complexity), n (Total operator + operand), total_Op (Total operator), branchCount, uniq_Opnd (Unique operands), lOCode (Count of Statement Lines), t (Time to write program), e (Effort to write program), uniq_Op (Unique operators), iv(g) (Design complexity), lOComment (Count of lines of comments) dengan akurasi 94,10 %. Pada dataset KC1 adalah fitur d (Difficulty) dengan akurasi 92,06%. Pada dataset KC2 adalah fitur d (Difficulty), dengan akurasi 96,65%. Pada dataset PC1 adalah fitur lOComment (Count of lines of comments),v (Volume), branchCount, v(g) (Cyclomatic complexity),lOCode lOCode (Count of Statement Lines), t (Time to write program), b (Error estimate), e (Effort to write program) dengan akurasi 93,02%.
ENGLISH:
Software defect or regularly known as bug or errors in aplications already created by a developer or programmer. Or Something that appears when the expectation from the software is not like with actual result. However, it can also when the testing finds something else from expectation, it’s not mean can be fully claimed as a bug. The data used is a dataset from NASA MDP, which amounts to 5 data, are CM1, JM1, KC1, KC2, and PC1.
This study performs feature selectin using Chi Square algorithm. The testing is done by selecting the features and then calculating the accuracy of prediction using Naive Bayes classification. with the take of different amounts on each data, it will be known later the number of features and features that have significant influence on this study.
The result of this study concluded the features that have a significant influence on the software defect prediction based on Chi Square feature selection on the CM1 dataset are features I (Intelligence), T (Time to write program) with accuracy 93.47%. Dataset JM1 are features I (Intelligence), L (Program length), total_Opnd (Total operand), D (Difficulty), V (Volume), B (Error estimate), loc (Line of Code), V (g) (Cyclomatic complexity), Ev (g) (Essential complexity), N (Total operator + operand), total_Op ( Total operator), branchCount, uniq_Opnd (Unique operands), lOCode (Count of Statement Lines), T (Time to write program), E (Effort to write program), uniq_Op (Unique operators), iv(g) (Design complexity), lOComment (Count of Lines of comments) with accuracy 94.10%. In the KC1 dataset is the feature D (Difficulty) with accuracy 92.06%. In the KC2 Dataset is the feature D (Difficulty), with accuracy 96.65%. In the PC1 dataset are features lOComment (Count of lines of comments), V (Volume), branchCount, V (g) (Cyclomatic complexity), lOCode lOCode (Count of Statement Lines), T (Time to write program), B (Error estimate), E (Effort to write program) with accuracy 93.02%.
Item Type: | Thesis (Undergraduate) | |||||||||
---|---|---|---|---|---|---|---|---|---|---|
Supervisor: | Fatchurrochman, Fatchurrochman and Santoso, Irwan Budi | |||||||||
Contributors: |
|
|||||||||
Keywords: | prediksi cacat perangkat lunak; seleksi fitur chi Square; software defect prediction; chi square feature selection | |||||||||
Departement: | Fakultas Sains dan Teknologi > Jurusan Teknik Informatika | |||||||||
Depositing User: | Heni Kurnia Ningsih | |||||||||
Date Deposited: | 30 Apr 2020 14:35 | |||||||||
Last Modified: | 30 Apr 2020 14:35 | |||||||||
URI: | http://etheses.uin-malang.ac.id/id/eprint/15307 |
Downloads
Downloads per month over past year
Actions (login required)
View Item |