Hasil Pencarian  ::  Simpan CSV :: Kembali

"In line with rapid business process digitalization in the Directorate General of Taxes, the size of the data stored in the institution has grown exponentially. However, there is a problem with generating value out of the valuable data assets. Correspondingly, this research provides machine-learning-based predictive analytics as a solution to the question of how to use taxpayers' trigger data as a decision support system to discover and realize unexplored tax potential. More specifically, this research presents predictive analytics models that can accurately predict which potential taxpayers are likely to pay their due. We developed three machine learning models: logistic regression, random forest, and decision tree. We analyzed 5,562 tax revenue potential data samples with eight predictors: trigger data nominal value, distance to tax office, type of taxpayer, media of tax report, type of tax, report status, registered year of taxpayer, and area coverage. Our study shows that the random forest model provided the best prediction performance. The resultant weight of each attribute indicated that the status of the tax report was the top tier of variable importance in predicting tax revenue potential. The analytics can help tax officers determine potential taxpayers with the highest likelihood to pay their due. Given the size of the data records, this approach can provide tax administrators with a powerful tool to increase work efficiency, combat tax evasion, and provide better customer service."

"[ABSTRAKPerkembangan sistem informasi saat ini menyebabkan sistem informasi yang digunakan dalam sebuah organisasi terus bertambah dan semakin kompleks. Halini juga memunculkan fenomena meningkatnya jumlah data yang diolah dandihasilkan oleh sistem informasi. Kondisi ini membawa tantangan baru dalampengawasan operasional sistem informasi, seperti keterlambatan peringatankesalahan atau membanjirnya jumlah peringatan yang tidak tepat sasaran. Penelitian ini bertujuan membangun sebuah sistem pengawasan aplikasi padasistem informasi di PT. XYZ menggunakan Event Driven Architecture dan Machine Learning. Pengembangan ini menggunakan perangkat lunak R dan TIBCO StreamBase.

---

ABSTRACTAdvancement in information system nowadays has generated more quantities and complexities of an organization?s information system. This factalso leads to a phenomenon of the increase of data volume being processed andalso generated by any information system. This condition has brought a newchallenge in the operation and monitoring of the information systems, such asdelays in failure alert and also floods of incorrect alerts. This research aims to build a monitoring system for applications in the PT. XYZ information systems, using Event Driven Architecture and Machine Learning techniques. This development is done using R software and also TIBCO StreamBase. , Advancement in information system nowadays has generated more quantities and complexities of an organization’s information system. This factalso leads to a phenomenon of the increase of data volume being processed andalso generated by any information system. This condition has brought a newchallenge in the operation and monitoring of the information systems, such asdelays in failure alert and also floods of incorrect alerts. This research aims to build a monitoring system for applications in the PT. XYZ information systems, using Event Driven Architecture and Machine Learning techniques. This development is done using R software and also TIBCO StreamBase. ]"

"Moda transportasi udara sangat bergantung pada keadaan dan perubahan cuaca, baik saat lepas landas, mengudara, maupun saat pendaratan, dengan lebih dari 50% kecelakaan pesawat diakibatkan oleh cuaca. Curah hujan yang tinggi dapat mengganggu aktivitas penerbangan dengan menurunkan visibilitas, meningkatkan massa pesawat, mengurangi akurasi instrumen pengukuran, serta menyebabkan turbulensi. Oleh karena itu, penting bagi manajemen bandara untuk memastikan kondisi cuaca aman bagi operasi pesawat. Penelitian ini bertujuan untuk mengembangkan model prediksi kategori hujan berdasarkan curah hujan untuk 1 jam, 3 jam, dan 9 jam ke depan, menggunakan data dari AWOS di Bandara Jenderal Ahmad Yani, Semarang. Algoritma yang digunakan adalah Random Forest dengan 100 pohon dan K-Nearest Neighbor (KNN) dengan k sebesar 5. Hasil analisis menunjukkan bahwa model KNN dan Random Forest memiliki performa yang cukup baik, dengan prediksi terbaik untuk periode 1 jam ke depan. Model KNN memiliki performa terbaik dengan akurasi 0,86, presisi 086, recall 0,86, F1-score 0,85, dan MCC 0,83.

---

Air transportation is highly dependent on weather conditions and changes, both during takeoff, flight, and landing, with more than 50% of aircraft accidents caused by weather. Heavy rainfall can disrupt flight activities by reducing visibility, increasing aircraft mass, decreasing the accuracy of onboard measurement instruments, and causing turbulence. Therefore, it is crucial for airport management to ensure that weather conditions are safe for aircraft operations. This study aims to develop a model to predict rain categories based on rainfall for 1 hour, 3 hours, and 9 hours ahead, using data from AWOS at Jenderal Ahmad Yani Airport, Semarang. The algorithms used are Random Forest with 100 trees and K-Nearest Neighbor (KNN) with k set to 5. The analysis results show that the KNN and Random Forest models perform reasonably well, with the best predictions made for the 1-hour ahead period. The KNN model demonstrated the best performance with an accuracy of 0.86, precision of 0.86, recall of 0.86, F1-score of 0.86, and MCC of 0.86."

"Just like electricity, Machine Learning will revolutionize our life in many ways-some of which are not even conceivable today. This book provides a thorough conceptual understanding of Machine Learning techniques and algorithms. Many of the mathematical concepts are explained in an intuitive manner. The book starts with an overview of machine learning and the underlying Mathematical and Statistical concepts before moving onto machine learning topics. It gradually builds up the depth, covering many of the present day machine learning algorithms, ending in Deep Learning and Reinforcement Learning algorithms. The book also covers some of the popular Machine Learning applications. The material in this book is agnostic to any specific programming language or hardware so that readers can try these concepts on whichever platforms they are already familiar with."

"Pembangunan Ibu Kota Negara yang terletak di Kalimantan Timur menjadi proyek besar pemerintah dengan anggaran biaya Rp 466 Triliun. Untuk menjamin efisiensi biaya konstruksi dari proyek ini, diperlukan pendekatan estimasi biaya yang baik. Namun estimasi biaya dalam konstruksi di Indonesia saat ini masih didominasi metode konvensional yang masih memiliki kekurangan dalam hal akurasi maupun waktu pemantauan. Oleh karena itu, penelitian ini dilakukan dengan tujuan untuk mengembangkan model machine learning untuk memprediksi estimate at completion biaya konstruksi sebagai alternatif dari metode konvensional. Penelitian ini dilakukan dengan metode analisis studi literatur dan benchmarking dari penelitian yang telah ada untuk menjalankan simulasi model menggunakan perangkat lunak RapidMiner yang kemudian akan divalidasi oleh narasumber pakar melalui wawancara. Diperoleh hasil bahwa model terbaik didapat menggunakan algoritma neural network. Dari simulasi model tersebut dengan menggunakan data dari salah satu proyek pembangunan jalan raya di Ibu Kota Negara, didapat hasil output berupa prediksi penghematan sebesar 17,8% dari nilai budget at completion proyek. Hasil dari prediksi model tersebut menghasilkan output yang lebih konservatif apabila diperbandingkan dengan metode konvension menggunakan formula estimate at completion biaya.

---

The development of the National Capital City located in East Kalimantan has become a major government project with a budget of IDR 466 trillion. To ensure cost efficiency in the construction of this project, a good cost estimation approach is required. However, cost estimation in construction in Indonesia is currently dominated by conventional methods that still have shortcomings in terms of accuracy and monitoring time. Therefore, this research is conducted with the aim of developing a machine learning model to predict the estimate at completion of construction costs as an alternative to conventional methods. The research is carried out through the analysis of literature studies and benchmarking from existing research to execute a simulation model using the RapidMiner software, which will then be validated by expert informants through interviews. The results indicate that the best model is obtained by using neural network algorithm. From the simulation model using data from one of the road construction projects in the National Capital City, the output shows a predicted savings of 17.8% from the project’s budget at completion value. This model prediction produce a more conservative result than conventional methods of estimating costs at completion."

"Machine learning focuses on predition-- using what you know to predict what you would like to know based on historical relationships between the two. At its core, it's a mathematical/algorithm-based technology that, until recently, required a deep understanding of math and statistical concepts, and fluency in R and other specialized languages. "Machine learning with Spark and Python" simplifies machine learning for a broader audience and wider application by focusing on two algorithm families that effectively predict outcomes, and by showing you how to apply them using the popular and accessible Python programming language. This edition shows how pyspark extends these two algorithms to extremely large data sets requiring multiple distributed processors. The same basic concepts apply. Author Michael Bowles draws from years of machine learning expertise to walk you through the design, construction, and implementation of your own machine learning solutions. The algorithms are explained in simple terms with no complex math, and sample code is provided to help you get started right away. You'll delve deep into the mechanisms behind the constructs, and learn how to select and apply the algorithm that will best solve the problem at hand, whether simple or complex. Detailed examples illustrate the machinery with specific, hackable code, and descriptive coverage of penalized linear regression and ensemble methods helps you understand the fundamental processes at work in machine learning. The methods are effective and well tested, and the results speak for themselves"

"Tingkat kemacetan di Jakarta saat ini tergolong tinggi dan memiliki tren yang meningkat setiap tahu. Terdapat berbagai upaya yang dilakukan oleh pihak manajemen kemacetan untuk mengurangi kemacetan. Salah satu komponen yang perlu diperhatikan pada perencanaan upaya dalam mengurangi kemacetan adalah penemuan atribut yang memiliki pengaruh kepada tingkat kemacetan. Pendekatan machine learning (ML) pada beberapa tahun terakhir memberi hasil yang baik berdasarkan nilai metrik performa model. Maka, penelitian ini menggunakan algoritma ML, yaitu support vector machine (SVM), k-nearest neighbors (KNN), dan random forest (RF) untuk membangun model dalam memprediksi kemacetan serta menemukan faktor yang memiliki pengaruh terhadap kemacetan di ruas jalan. Variabel independen yang digunakan pada penelitian ini adalah jam, hari kerja, tanggal merah, curah hujan, ada tidaknya event, jam ganjil genap, volume motor, volume mobil, serta volume bus dan truk. Variabel dependen yang digunakan adalah tingkat kemacetan yang mewakili kecepatan rata-rata kendaraan di ruas jalan. Model dijalankan pada dua data, yakni pada data dengan variabel volume kendaraan dan data tanpa variabel kendaraan. Hasil penelitian menunjukkan model SVM, KNN, dan RF memberikan nilai akurasi, precision, recall, dan F1 score di atas 80% pada kedua data. Adapun faktor yang memiliki pengaruh kuat terhadap tingkat kemacetan terdiri dari jam dan jam ganjil genap pada data tanpa volume kendaraan serta volume motor, volume mobil, volume bus dan truk, jam, dan jam ganjil genap pada data dengan volume kendaraan.

---

The level of congestion in Jakarta is currently high and has an increasing trend every year. There are various efforts made by congestion management to reduce congestion. One component that needs to be considered in planning efforts to reduce congestion is the discovery of attributes that have an influence on the level of congestion. Machine learning (ML) approaches in recent years have provided good results based on the value of model performance metrics. So, this study uses ML algorithms, namely support vector machine (SVM), k-nearest neighbors (KNN), and random forest (RF) to build a model to predict congestion and find factors that have an influence on congestion on road sections. The independent variables used in this study are hours, weekdays, red dates, rainfall, presence or absence of events, even odd hours, motorcycle volume, car volume, and bus and truck volume. The dependent variable used is the level of congestion, which represents the average speed of vehicles on the road. The model was run on two data, namely on data with vehicle volume variables and data without vehicle variables. The results showed that the SVM, KNN, and RF models provided accuracy, precision, recall, and f1 score values above 80% on both data. The factors that have a strong influence on the level of congestion consist of hours and even odd hours on data without vehicle volume and motorcycle volume, car volume, bus and truck volume, hours, and even odd hours on data with vehicle volume."

"Memprediksi niat kunjungan kembali memainkan peran penting dalam kebangkitan kembali waktu pandemi yang akan menguntungkan keunggulan kompetitif jangka pendek dan jangka panjang. Penelitian ini mengkaji faktor-faktor penentu niat berkunjung kembali dari analisis sentimen berbasis aspek dan pembelajaran mesin. Pendekatan big data diterapkan pada empat set data atraksi, hotel bintang 4&5, hotel bintang 3, dan motel dengan 49.399 ulasan dari TripAdvisor. Kami menerapkan metode pemodelan topik untuk mengekstrak aspek dan atribut, menghasilkan 10 aspek untuk kategorisasi hotel 4&5 dan kumpulan data atraksi, 6 aspek pada kumpulan data hotel bintang 3 dan Motel. Hasil analisis sentimen menunjukkan bahwa sentimen wisatawan secara positif dan negatif juga mempengaruhi kemungkinan niat berkunjung kembali. Peneliti menerapkan metode Logistic Regression, Random Forest Classifier, Decision Tree, k-NN, dan XGBoost untuk memprediksi niat berkunjung kembali yang menghasilkan tiga topik utama yang mendominasi probabilitas niat berkunjung kembali untuk masing-masing dataset. Aspek Properti pada hotel bintang 4&5 dan hotel bintang 3 mengindikasikan memiliki kemungkinan tinggi untuk niat berkunjung kembali. Sedangkan aspek Motels pada Atmosfir, Aktivitas Wisata, dan Durasi cenderung memiliki probabilitas niat berkunjung kembali. Aspek atraksi pada Harga, Layanan, Suasana meningkatkan kemungkinan niat berkunjung kembali. Studi ini berkontribusi pada pemanfaatan data besar dan pembelajaran mesin di industri pariwisata dan perhotelan dengan berfokus pada strategi inovatif sebagai pengurangan biaya untuk mempertahankan niat kunjungan kembali di kebangkitan kembali dari pandemi.

---

Predicting revisit intention plays a crucial role in the reawakening time of pandemic that will benefit short-term and long-term competitive advantage. This study examines the determiner factors of revisit intention from aspect-based sentiment analysis and machine learning. A big data approach was applied on four datasets of attractions, hotel 4&5 stars, hotel 3 stars, and motels with 49,399 reviews from TripAdvisor. We applied a topic modeling method to extract aspects and attributes, resulting in 10 aspects for hotel 4&5 categorization and attractions dataset, 6 aspects on hotel 3 stars and Motels dataset. Results on sentiment analysis show that tourists’ sentiment in positives and negatives also affect probability of revisit intention. Researchers applied methods of Logistic Regression, Random Forest Classifier, Decision Tree, k-NN, andXGBoost to predict revisit intention resulting in three main topics that have dominated probability on revisit intention for each dataset respectively. Aspect Properties on hotels 4&5 stars and hotel 3 stars indicate to have a high probability of revisit intention. Meanwhile, Motels' aspects on Atmosphere, Tourist Activities, and Duration tend to have a probability of revisit intention. Attraction’s aspects on Price, Services, Ambience increase probability of revisit intention. This study contributes to the utilization of big data and machine learning in tourism and hospitality industry by focusing on an innovative strategy as cost reduction to maintain revisit intention in the reawakening from pandemic."

"Dalam menghadapi meningkatnya permintaan transportasi muatan curah kargo di era transisi energi, efisiensi operasional menjadi krusial untuk mengelola biaya operasional harian kapal bulk carrier, di mana bahan bakar mencakup 60%-70% dari total biaya operasional. Penelitian ini mengusulkan pendekatan menggunakan model Random Forest (RF) untuk memprediksi konsumsi bahan bakar kapal, mengatasi keterbatasan metode empiris statistik konvensional dalam memodelkan faktor eksternal seperti kondisi cuaca. Ordinary Least Squares (OLS) digunakan untuk mengevaluasi signifikansi variabel independen setelah normalisasi data dengan metode min-max, dengan pembagian data training dan testing sebesar 70% dan 30%. Pendekatan baru diterapkan untuk validasi data guna mengevaluasi sejauh mana model dapat membaca dataset dengan variasi jumlah subset data kapal, dan menggunakan analisis histogram untuk mengkaji pergeseran nilai error dalam persebaran data seiring bertambahnya jumlah data yang digunakan. Evaluasi dilakukan menggunakan empat metrik, yaitu MSE, RMSE, MAE, dan MAPE, yang menunjukkan bahwa model RF mencapai akurasi tinggi sebesar 95%-98% dengan kesalahan rata-rata sangat rendah di bawah 0,1 pada semua metrik. Penelitian ini tidak hanya memberikan solusi efektif untuk mengoptimalkan konsumsi bahan bakar dan meminimalkan biaya operasional, tetapi juga mendukung pengambilan keputusan yang lebih cepat dan tepat dalam operasional kapal.

---

In response to the increasing demand for bulk cargo transportation in the energy transition era, operational efficiency is crucial to managing the daily operational costs of bulk carrier vessels, with fuel accounting for 60%-70% of total operational expenses. This study proposes an approach utilizing the Random Forest (RF) model to predict ship fuel consumption, addressing the limitations of conventional empirical statistical methods in modeling external factors such as weather conditions. Ordinary Least Squares (OLS) was employed to evaluate the significance of independent variables after data normalization using the min-max method, with a 70% and 30% split for training and testing data, respectively. A novel approach was implemented for data validation to assess the extent to which the model can interpret datasets with varying subsets of ship data, using histogram analysis to examine the shift in error distribution as the dataset size increases. The evaluation was conducted using four metrics, namely MSE, RMSE, MAE, and MAPE, demonstrating that the RF model achieved high accuracy between 95% and 98%, with extremely low average errors below 0.1 across all metrics. This study not only provides an effective solution to optimize fuel consumption and minimize operational costs but also supports faster and more accurate decision-making in ship operations."