Claim Missing Document
Check
Articles

Optimalisasi Model Logistic Regression untuk Prediksi Diabetes Menggunakan Seleksi Fitur Berbasis Korelasi Nugraha, Wahyu; Syarif, Muhamad
Jurnal ICT: Information Communication & Technology Vol. 25 No. 2 (2025): JICT-IKMI, December , 2025
Publisher : LPPM STMIK IKMI Cirebon

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.36054/jict-ikmi.v25i2.321

Abstract

Diabetes Mellitus is a pressing global health challenge, making early detection a key component of effective intervention. Machine learning has shown great potential in predicting diabetes risk. Among various models, Logistic Regression (LR) is often favored in a medical context due to its high interpretability, although its accuracy frequently lags behind more complex black-box models. LR performance is known to be highly sensitive to the quality and relevance of input features. This study aims to quantitatively evaluate the impact of a strict correlation-based feature selection strategy on the accuracy of the Logistic Regression model. Using the "Diabetes Health Indicators" dataset (N=100,000), this study compares two scenarios: (1) a baseline LR model using all features (All Input) and (2) an optimized LR model using only a subset of features (including engineered features) that have a high absolute correlation with diabetes diagnosis (Correlated Input). The results demonstrate a significant performance improvement. The All Input baseline model achieved an accuracy of 80.45%, while the Correlated Input model achieved an accuracy of 85.67%. This +5.22% absolute increase demonstrates that correlation-based feature selection effectively eliminates noise from irrelevant features, thus drastically improving the predictive power of the LR model. This study concludes that an optimized Logistic Regression with feature selection offers a strong balance between improved accuracy and the interpretability essential for clinical applications.
Meningkatkan Daya Saing UMKM Desa Punggur Besar Melalui Strategi Pemasaran Digital Berbasis Kecerdasan Buatan (AI) Anna, Anna; Annisa, Riski; Rahayuningsih, Panny Agustia; Nugraha, Wahyu
Mestaka: Jurnal Pengabdian Kepada Masyarakat Vol. 5 No. 1 (2026): Februari 2026
Publisher : Pakis Journal Institute

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.58184/mestaka.v5i1.838

Abstract

The Free Nutritious Meals Program is a social initiative managed by micro, small, and medium enterprises (MSMEs) in Punggur Besar Village to support the nutritional needs of schoolchildren. However, program managers face challenges in increasing visibility, transparency, and community engagement due to limited digital technology capabilities. This activity aims to empower program managers through AI-based digital marketing training that is easily accessible and relevant to their social context. The training was conducted offline at the Punggur Besar Village Office using participatory learning methods and hands-on practice using participants' own devices. Twenty program managers participated in the entire activity. All participants successfully created program-specific social media accounts, produced educational content, and documented activities with the help of AI tools. Eighty percent of participants were able to independently develop content narratives, and seventy-two percent began utilizing basic analytics to understand community responses to their content. The main benefits gained were increased digital communication capacity, strengthened program accountability, and increased confidence in interacting with the public online. This activity demonstrated that the use of AI in digital marketing can be adapted inclusively to strengthen community-based social programs, with the active involvement of managers as key actors in digital transformation at the village level.
ANALISIS BUTIR SOAL PENILAIAN MATA PELAJARAN EKONOMI DALAM KAITANNYA DENGAN ASPEK KOGNITIF TAXONOMY BLOOM Nugraha, Wahyu; Harini, Harini; Sudarno, Sudarno
BISE: Jurnal Pendidikan Bisnis dan Ekonomi Vol 2, No 2 (2016): Jurnal Pendidikan Bisnis dan Ekonomi
Publisher : Department of Economics Education, Universitas Sebelas Maret

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.20961/bise.v2i2.16872

Abstract

This research aims to identify the quality of items about the assessment of subjects Economics class XI in SMA Negeri 4 Surakarta in the academic year  2016/2017 in relation to cognitive aspects Taxonomy Bloom. The subjects chosen are Economics subject teachers and students of XI IPS class in SMA Negeri 4 Surakarta academic year 2016/2017, the subject taking technique in this study using the posposive sampling. Meanwhile, the technique used in sampling the data source is by using snowball sampling technique. This research is a descriptive research with qualitative approach. Data collection was obtained by observation method for test development activity data, interview method for test development data and quality of item and documentation for data item quality. The results showed that: (1) a valid question a total of 11 items (36.7%) while invalid 19 items (63.3%). (2) Based on the reliability, including the matter of which the reliability is very low, -0.057. (3) Based on the level of difficulty, including difficult level items totaled 25 questions (83%), moderate level 2 questions (7%), and an easy level 3 questions (10%). (4) Based on the distinguishing features, items that includes not very good 5 questions (16.7%), there is a good 5 questions (16.7%), 10 items are good enough (33.3%), 8 points are good (26, 7%), and splendidly questions are 2 points (6.67%). (5) Based on the effectiveness of the use of distractors, items which serve very good 3 point (10%), serves as good 8 points (26.7%), and 10 items as enough (33.3%), less 5 questions (16.7%), and serves no good / bad 4 items (13.3%). (5) Based on the analysis validity, level of difficulty, distinguishing, and the effectiveness of the use of distractors there are good-quality questions amount to 2 questions (6.7%), unfavorable 6 matter (20%), and not good or bad 22 questions (73.3%). Based on the cognitive aspects of Taxonomy Bloom, the majority of items are dominated by items with category C1 consisting of 13 items (43.3%), category C2 amounted to 11 items (36.7%), category C3 amounted to 3 items (10%) and the items with category C4 amounted to 3 items (10%). Overall matter of assessment of subjectsEconomics class XI SMA Negeri 4 Surakarta Academic year 2016/2017 in relation to the cognitive aspects of Taxonomy Bloom including the problem that is not good.