EXPERT
Not a member yet
201 research outputs found
Sort by
The Comparison of Brute Force, Cheapest-Insertion, and Nearest-Neighbor Heuristics for Determining the Shortest Tour for Visiting Malls in Bandar Lampung
Full text linkA mall is a sizable retail complex made up of numerous stores, as well as restaurants and other commercial spaces, all situated in one huge building or a collection of related or neighboring buildings. In this study we will compare the heuristics which are Brute Force, Cheapest-Insertion, and Nearest-Neighbor heuristics to find the shortest route from one mall to other eight malls in Bandar Lampung exactly once and back to the origin. The problem of finding the shortest route from original location, and go to every location exactly once, and back to original location is known as The Travelling Salesman Problem. Implementing on the data of the locations of those nine malls, the Brute Force Heuristic gives the total length 32,8 km, Cheapest-Insertion Heuristic 33,2 km, and Nearest-Neighbor Heuristic 33,35 km. However, in term of efficiency, the Brute Force is not efficient because we have to search for all possible solutions
Analisis Performa Deteksi Cacar Monyet dengan Model Klasifikasi Gambar Menggunakan Teachable Machine dan Keras
Full text linkDigital Image Detection of Monkeypox Disease Performance Analysis using Image Classification Model with Teachable Machine and Keras. Advancements in artificial intelligence technology, particularly in the field of machine learning, have opened up significant new opportunities for developing disease detection systems that are faster, more accurate, and more efficient. This study aims to analyze the performance of Teachable Machine and Keras models in classifying monkeypox images using a quantitative approach. A dataset of skin images with indications of monkeypox was collected for the development of image classification models using both tools. The results of the study show that the developed models accurately recognized images that had been seen during the training data for the "Monkey Pox" category. However, when faced with test images that had not been seen before, the models showed limitations in generalizing, indicating overfitting in that category. Conversely, for the "Other" category, the models were able to recognize well both in the training data and the test data, demonstrating better generalization capability in this category. Therefore, for future research, it is recommended to conduct a more in-depth evaluation of the use of Data Augmentation techniques to expand data variation, as well as to explore other platforms or tools that can provide greater control over test data management
The Prediction of Subsidized Fertilizer Stock Using Least Square Support Vector Machine on The Kartu Petani Berjaya Aplication
Full text linkAgriculture is one of the biggest commodities in Lampung, so that this also causes a lot of use and allocation of subsidized fertilizers. In terms of this it is very important to know how much amount of subsidized fertilizer needed in the future to prepare subsidized fertilizer stocks. The data needed was the time series data from subsidized fertilizer redemption data, using Least Square Support Machine and Autoregressive Integrated Moving Average methods to make a prediction model for subsidized fertilizer redemption. The result was hoped that we can find out how many harvests are in Lampung and the future subsidized fertilizer rations. This research was expected to provide benefits to the relevant parties
Analisis Performansi Naïve Bayes pada Klasifikasi Plagiarisme Dokumen Berdasarkan Pembobotan Teks Menggunakan Algoritma TF-IDF
Full text linkProposal skripsi adalah rencana penelitian yang diajukan oleh mahasiswa dengan bimbingan dari Dosen Pembimbing, dan disusun mengikuti aturan penulisan karya ilmiah. Algoritma TF-IDF adalah metode statistik numerik yang menunjukkan seberapa penting suatu kata dalam sebuah dokumen atau korpus. Metode Naive Bayes Classifiers adalah teknik klasifikasi teks yang menggunakan probabilitas kata kunci untuk membandingkan dokumen pelatihan dengan dokumen uji. Penelitian ini bertujuan untuk melakukan analisis performansi Naïve Bayes Classifier pada klasifikasi plagiarisme dokumen proposal skripsi berdasarkan pembobotan teks menggunakan algoritma TF-IDF. Analisis performansi dilakukan menggunakan metode Confusion Matrix. Berdasarkan pengujian dan analisis yang dilakukan menggunakan uji performansi Naive Bayes Classifier untuk klasifikasi kelas Plagiarisme Rendah, Plagiarisme Sedang, dan Plagiarisme Berat dengan jumlah dataset 85 dokumen proposal skripsi yang terbagi dalam Data Training dan Data Testing. Jumlah Data Training berjumlah 59 korpus dan Data Testing berjumlah 26 korpus. Berdasarkan uji performansi yang dilakukan mengguanakn metode Confusion Matrix didapatkan hasil yang ditunjukan pada Tabel 6 dengan Split Data 70 : 30, dengan nilai Accuracy 97,65%, nilai Precision 95,23%, dan nilai Precall 98,74%. Hal ini menunjukkan bahwa Naive Bayes Classifier berada pada tingkat excellent classification. Untuk penelitian berikutnya, analisis kritis lebih tinggi dalam dataset, maka prediksi pada data testing semakin akurat. Dengan precision sebesar 95,23%, recall sebesar 98,74%, dan accuracy sebesar 97,65%, dapat disimpulkan bahwa algoritma Naïve Bayes Classifier menunjukkan tingkat excellent classification
Aplikasi Pencatatan Keuangan Berbasis Web mobile Pada Unit Bisnis Instidla
Full text linkDengan berkembangnya teknologi informasi, khususnya dalam era digital yang terus berlangsung, pembaruan dalam pencatatan seharusnya mampu mencerminkan kemajuan dalam pengelolaan data. Salah satu permasalahan yang dihadapi oleh Unit Bisnis Instidla adalah metode pencatatan keuangan yang masih ditulis dalam bentuk penulisan di dalam buku laporan. Oleh karena itu, penelitian ini diarahkan untuk mengembangkan aplikasi sistem informasi Pencatatan Keuangan Guna Menjaga kerentanan akan kehilangan data dikarenakan buku hilang atau rusak dan diharapkan dapat membantu dalam pengelolaan pencatatan keuangan yang akurat, sistem informasi ini dirancang menggunakan teknologi Progressive Web App, mempermudah akses tanpa memerlukan instalasi aplikasi terlebih dahulu, serta dapat diakses secara online maupun offline. Dengan demikian, aplikasi ini diharapkan dapat menjadi solusi bagi Unit Bisnis Instidla dalam mengatasi tantangan dalam pengelolaan keuangan. Melalui metode RnD dan pendekatan ADDIE, penelitian ini mencakup analisis kebutuhan, desain antarmuka, pengembangan aplikasi sesuai kebutuhan Unit Bisnis di Instidla, Implementasi sistem, dan evaluasi. Hasil dari penelitian ini adalah rancangan aplikasi yang berfokus pada fungsi pencatatan keuangan yang akurat, yang bertujuan memberikan solusi praktis untuk meningkatkan kinerja pencatatan keuangan di Unit Bisnis Instidla. Hasil pengujian menunjukkan bahwa fungsi pencatatan keuangan pada aplikasi dapat berjalan dengan baik dan sesuai dengan spesifikasi kebutuhan Unit Bisnis Instidla
Estimasi Proyek Aplikasi Online Shop dengan COCOMO II Menggunakan Pendekatan Algoritma SPRINT
Full text linkSoftware Project Estimation for an Online Shop Application Using COCOMO II with the SPRINT Algorithm. Effort estimation is an important factor in successful software development. COCOMO II is an estimation model that can estimate application project effort using scale factors and cost drivers. However, the accuracy of this model is considered low, so this research aims to improve the estimation accuracy of the COCOMO II model by applying the SPRINT algorithm approach to the research object of the online shop application project. The COCOMO II model is used to calculate the estimation of project time, personnel, and costs. Meanwhile, the SPRINT algorithm is used to predict the priority of module work based on the COCOMO II estimation results. This study compares the accuracy of effort estimation with previous research using the COCOMO II model with the C4.5 algorithm approach which is the predecessor of the SPRINT algorithm. The results show that the estimation of application projects using the COCOMO II model with the SPRINT algorithm approach produces 100% accuracy, more accurate than the COCOMO II model with the C4.5 algorithm approach which only produces 90% accuracy. This research proves that the use of the SPRINT algorithm can further improve the speed and accuracy of prediction compared to the use of the C4.5 algorithm
Implementasi dan Analisis Akurasi Pengukuran Luas Wilayah Kota Bekasi Menggunakan Algoritma Divide dan Conquer & Metode Grid
Full text linkImplementation and Accuracy Analysis of Area Measurement in Bekasi City Using Divide and Conquer Algorithm & Grid Method. Measuring area is an important factor in an urban planning of an area. This is because area size is often used as a budget basis in various development projects, especially those in physical form. Accurate area measurement will certainly have an impact on the budget policies that will be taken. One method that can be used to measure area is by applying the divide & conquer algorithm and the Grid method. The divide & conquer algorithm is an algorithm that divides complex/large problems into smaller sub-problems. Grid method is one kind of this algorithm. This method was chosen because it is simple, easy to understand, and provides quite optimal results. In the implementation of the Grid method, the map of the Bekasi City area will be divided into smaller areas and mapped them in Grid form. The calculation results show quite good results, that is the area of Bekasi City is 208 km2. Compared to the actual area of 210,49 km2, so the difference area size is 2,49 km2. So the accuracy level of using this method is 98,82%
Teknik Prediksi Data Mining pada Perguruan Tinggi sebagai Kajian Literatur
Full text linkHigher education is an interesting research object because there are complex research topics that can be used by students, lecturers or researchers. One of the research topics that uses university research objects is related to data mining. Data mining is a technique that can be used to extract data into knowledge and functions to find patterns from large data. There are currently 5 data mining roles used, namely Estimation, Prediction, Classification, Clustering and Association. One of the techniques used is Prediction. The aim of this research is to conduct systematic literature review research related to data mining prediction techniques with 3 aspects, namely Algorithms, Frameworks/Methods and Research Topics. The results of this research are the Algorithm Trends used in the research, namely the Support Vector Machines (SVM), Random Forest, Decision tree (DT), Artificial Neural Networks (ANN), Naive Bayes (NB), Neural Network (NN), and K- Nearest Neighbor (KNN) Framework used is Data Mining, Educational data mining and Machine Learning and the trending topic is related to Prediction, namely measuring Student performance
Analisis Perancangan Sistem Informasi Portal Desa Berbasis Web dengan Menggunakan Metode Kano dan OpenSID
Full text linkAnalysis and Design of Web-Based Village Information System Using Kano and OpenSID Method. The internet provides a very important role in the process of distributing information, as well as village information must be immediately distributed to the community so that village activities can run well and successfully. But the problem is because the information conveyed by the village is still conventional, many people do not know so that the information that people should know does not know and results in village development not running smoothly. From this, there is a need for innovation in the delivery of village information, such innovations include a village information system that can be accessed by villagers anytime and anywhere. OpenSID is a village information system platform that provides features including contains information about the village, health information, village identity, population data, population data statistics, correspondence, public information, village finances, village data analysis, government assistance programs, land data, village maps and facilities for complaints and criticism of the 12 features, a needs analysis was carried out using the Kano method whose data was taken from 126 respondents with the results of the features entered into Must be (compulsory features) were Health Information, Correspondence, Public Information, Village Finances, Village Maps and Facilities for Submitting Complaints and Criticisms. Then, based on the Non-Parametric Statistics test (Kendall’s Tau-b) of the Kano Method in extracting user needs in the Village information system, it was found that all pairs of features had the same direction of correlation because all the correlation coefficients were positive. The resulting correlation is significant at the 99% confidence level. However, the correlation coefficient value is not too large and is in the range of 0.213-0.553, where the strength of the correlation ranges from weak to moderate
Aplikasi Point of Sale untuk Meningkatkan Profitabilitas dan Digitalisasi UMKM
Full text linkMenurut data dari Kementerian Koperasi dan UKM, pada tahun 2020 terdapat sekitar 64,2 juta UMKM yang tersebar di seluruh Indonesia termasuk di Kabupaten Pringsewu Provinsi Lampung. Mayoritas usaha kecil dan menengah (UMKM) di Kabupaten Pringsewu masih menggunakan metode kuno untuk mencatat transaksi menggunakan kertas, terutama di toko kelontong yang masih menggunakan proses manual. Pencatatan manual dengan buku kertas memiliki banyak risiko, mulai dari kerusakan fisik hingga kehilangan data karena kecelakaan atau pencurian. UMKM juga menghadapi masalah bagaimana mengintegrasikan semua cabang bisnis mereka, seperti yang terjadi pada toko kelontong yang juga mengembangkan bisnis pembuatan kue basah atau minuman jus. Tujuan dari penelitian ini adalah untuk membuat rancangan aplikasi Point of Sale (POS) Multi-Cabang yang diharapkan dapat meningkatkan profitabilitas dan digitalisasi UMKM Kabupaten Pringsewu. Hasil dari penelitian ini adalah aplikasi Point Of Sale Multi-Cabang yang diberinama UMKM Maju dengan fitur-fitur seperti transaksi penjualan barang, pengelolaan barang masuk, laporan penjualan barang, laporan pembelian barang, pengelolaan stok barang, pengelolaan cabang, pengelolaan pengguna aplikasi. Berdasarkan hasil evaluasi sistem menunjukan aplikasi dapat berjalan di berbagai platform, dan fungsionalitas aplikasi sudah sesuai dengan kebutuhan fungsional pelaku UMKM Kabupaten Pringsewu