Jurnal Informatika: Jurnal Pengembangan IT
Not a member yet
437 research outputs found
Sort by
A Comparative Study of MobileNet Architecture Optimizer for Crowd Prediction
No full textArtificial intelligence technology has grown quickly in recent years. Convolutional neural network (CNN) technology has also been developed as a result of these developments. However, because convolutional neural networks entail several calculations and the optimization of numerous matrices, their application necessitates the utilization of appropriate technology, such as GPUs or other accelerators. Applying transfer learning techniques is one way to get around this resource barrier. MobileNetV2 is an example of a lightweight convolutional neural network architecture that is appropriate for transfer learning. The objective of the research is to compare the performance of SGD and Adam using the MobileNetv2 convolutional neural network architecture. Model training uses a learning rate of 0.0001, batch size of 32, and binary cross-entropy as the loss function. The training process is carried out for 100 epochs with the application of early stop and patience for 10 epochs. Result of this research is both models using Adam's optimizer and SGD show good capability in crowd classification. However, the model with the SGD optimizer has a slightly superior performance even with less accuracy than model with Adam optimizer. Which is model with Adam has accuracy 96%, while the model with SGD has 95% accuracy. This is because in the graphical results model with the SGD optimizer shows better stability than the model with the Adam optimizer. The loss graph and accuracy graph of the SGD model are more consistent and tend to experience lower fluctuations than the Adam model
Analisis Crowdsourcing Dari Media Sosial Untuk Sumber Data Alternatif Kajian Bentang Lahan
No full textPeningkatan jumlah pengguna media sosial dari waktu ke waktu memberikan potensi baru dalam akuisisi data crowdsourcing. Proses akuisisi data tidak lagi membutuhkan banyak biaya dan waktu, karena crowdsourcing dapat digali dengan mudah bahkan tanpa biaya. Kajian ini mengangkat permasalahan apakah data crowdsourcing dari media sosial dapat dijadikan data alternatif dalam kajian geo-informatika. Proses akuisisi data dilakukan dari unggahan pengguna di media sosial. Unggahan menyertakan titik lokasi dan teks pada keterangan. Data yang diperoleh kemudian diolah untuk mengetahui titik lanskap dan kecenderungan penggunaan bahasa. Penggunaan bahasa dianalisis dengan metode RQDA dan diperoleh hasil 5,37% berbicara tentang bentang alam. Sedangkan titik lokasi media sosial dibandingkan dengan data DEMNAS memiliki skor akurasi 437,8 yang divalidasi dengan metode RMSE dan tidak direpresentasikan mendekati 1,0. Disarankan bahwa data media sosial masih jauh untuk dapat menjadi alternatif sumber data lanskap
Klasifikasi Jamur Berdasarkan Genus Dengan Menggunakan Metode CNN
No full textMushrooms are plants that do not have true roots and leaves. There are many types of mushrooms that have been identified worldwide, with various shapes, sizes, and colors. Mushrooms have many benefits in the fields of economy, health, and others. One of the benefits of mushrooms is as a food source in Indonesia, but not all types can be consumed. To identify mushroom species, the concepts of Genus and species can be used. The concept of Genus is considered easier because it groups mushroom types based on similar morphological characteristics. Therefore, a model is needed to classify mushrooms based on consumable and toxic genera. The method used in this research is Convolution Neural Network (CNN) due to its good predictive results in image recognition. The model in the research utilizes three convolution layers, three MaxPooling layers, and two dropout layers. The use of dropout aims to reduce overfitting in the model. The research uses a dataset of 1200 images with a training and testing data ratio of 70:30, resulting in 840 training data and 360 testing data. The best accuracy achieved by this model is 89% for training and 82% for validation. Therefore, it can be concluded that the model is able to classify mushrooms based on Genus using the CNN metho
Klasifikasi Nasabah Potensial menggunakan Algoritma Ensemble Least Square Support Vector Machine dengan AdaBoost
No full textIn the era of business and economics that are interconnected with each other and competition between companies in seeking market share so that there will be an increase, especially in the number of customers, especially deposit customers, financial institutions and other companies are increasingly realizing the importance of understanding and identifying potential customers correctly to get potential customers. customers subscribe to deposits. Potential customer classification is a strategic approach that allows financial institutions to identify potential customers who have the potential to subscribe to deposits. With a deeper understanding of the characteristics and needs of potential customers, financial institutions can direct marketing resources more effectively, increase marketing efforts, and increase the conversion of potential customers to active customers. The aim of this research is to develop and test the Ensemble Least Square Support Vector Machine model with AdaBoost in classifying potential customers which can increase accuracy in identifying potential customers who have the potential to subscribe to deposits. The research results showed that this method achieved an accuracy of 95.15%, a sensitivity of 92.93%, and a specificity of 97.61%. In comparison with single Support Vector Machine and Least Squares Support Vector Machine models, the Ensemble Least Squares Support Vector Machine outperforms both in terms of accuracy
Studi Komparasi Algoritma SVM Dan Random Forest Pada Analisis Sentimen Komentar Youtube BTS
No full textSentiment analysis of YouTube boy group BTS comments uses the NLP approach to detect emotional patterns based on two category labels, namely positive and negative. With NLP, positive or negative polarity in an entity can be allocated as well as predicted high and low performance from various classification sentiments. The machine learning algorithms used to measure the accuracy of sentiment analysis developed are the Support Vector Machine and Random Forest algorithms. The steps taken start from the data collection obtained from the BTS YouTube Comment dataset and then go through the data preprocessing stage. Then proceed to the feature extraction stage by converting text into digital vectors or Bag of Words (BOW) and classified using machine learning algorithms until the evaluation stage. From the results comparison of the evaluated algorithms, the accuracy value between the two algorithms is 96% for training data and 85% for data testing using the SVM algorithm, while for the Random Forest algorithm it is 82% for training data and 80% for data testing. This shows that the SVM algorithm produces a higher accuracy value than the Random Forest for sentiment analysis of YouTube boy group BTS comments
Pemanfaatan Narrowband IoT (NB-IoT) dalam Peningkatan Produktivitas Peternakan melalui Monitoring Otomatis
No full textThe rapid advancements in Narrowband IoT (NB-IoT) technology present significant opportunities for creating innovative products that can be implemented in daily life. One of these innovative products is the utilization of NB-IoT for monitoring cage conditions, maintenance, and boosting livestock productivity under challenging conditions that are difficult to manually control. This study aims to design an automated system capable of maintaining ideal cage conditions, including temperature, humidity, levels of ammonia (NH3 and CO2), as well as providing feed/water to livestock automatically and periodically. The research methodology involves the integration of various sensors mounted on a microcontroller, such as temperature sensors, humidity sensors, ammonia sensors, water level sensors, and pH sensors. The program executed by this microcontroller is connected to a control panel, and through the internet network, control and monitoring can be carried out using mobile and desktop devices. The test results indicate that this system is capable of providing ease in controlling the chicken coop for owners and workers, maintaining poultry health, and increasing livestock product yields from 97.17% of harvested poultry to 98.263%, with a decrease in the mortality rate from 2.830% to 1.737%. Overall, the utilization of NB-IoT technology in this research provides a positive impact on livestock management, offering an automated solution that enhances efficiency and productivity in the agricultural sector
Analisis Sentimen Fenomena PHK Massal Menggunakan Naive Bayes dan Support Vector Machine
No full textTermination of employment (PHK) on a large scale has a very significant impact on society and the economy. Mass layoffs have led to an increase in the number of unemployed people. Many people who have lost their jobs without a stable source of income struggle to find new jobs. This exacerbated the situation on the labor market and increased the number of unemployed people. Mass layoffs can also reduce economic activity and consumption. The sentiment analysis carried out aims to determine public sentiment regarding the phenomenon of mass layoffs that are currently happening in Indonesia based on positive and negative categories. In this study, the classification method used is the SVM method, which is one of the supervised learning methods in machine learning and also uses Nave Bayes as a comparison method. After classification, the next stage is the testing process using the K-fold cross-validation method. From the various sentiments obtained from Twitter data, it can be concluded that there are around 108 positive sentiments and 333 negative sentiments related to mass layoffs, while the results obtained from the test results using the SVM method show an accuracy of up to 84% while using the Nave Bayes method shows an accuracy of up to 74.1 percen
Visualisasi Data Analisa Sentimen RUU Omnibus Law Kesehatan Menggunakan KNN dengan Software RapidMiner
No full textThe government's decision to discuss the RUU Omnibus law on health has become a controversial topic in society, especially among users of the Twitter social media platform. Users express their opinions regarding their stance on the RUU Omnibus law through tweets on Twitter. With diverse comments from users, it is essential to classify and visualize them into useful information about the positive and negative sentiments towards the RUU on health. This is crucial to understand the public's response to this policy. A total of 2406 sentiment data from Twitter users were collected using the RapidMiner software. Before analyzing the data using the K-Nearest Neighbors (KNN) algorithm, data preprocessing was carried out. After preprocessing, 2.406 data points were obtained, which were then divided into 1.684 tweets for testing and 722 tweets for training. The data was then processed using the KNN algorithm model executed in the RapidMiner software. The results of the data processing were presented in the form of tables, graphs, and word clouds, aligning with the research objective of providing clear and easily understandable visualizations about the RUU on health. This facilitates understanding for stakeholders without technical backgrounds to grasp the meaning and sentiments expressed. The research results indicate that the testing of K-Nearest Neighbors (KNN) yielded a high accuracy value, making it well-visualized at 84.58%. This indicates that the KNN model is highly successful in analyzing Twitter users' opinions on the Health Omnibus Law based on the data used and its ability to visualize effectivel
Klasifikasi Citra Virus SARS-COV Menggunakan Deep Learning
No full textVarious variants of the SARS-COV virus emerged from 2003 to early 2022. This resulted in a heavy burden on the health sector in carrying out its duties and public services. It would be very helpful if a computer-assisted application was available that could distinguish between the variants of the SARS-CoV virus. From a scientific point of view, this is an opportunity for information technology to play its role to classify SARS-COV variants using supporting algorithms, including the use of artificial intelligence. Artificial intelligence-based and computer-assisted processes are certainly more immune to negative effects due to repetitive works and fatigue. In this study, Classification of the SARS-COV Virus Image Using Deep Learning (CNN) was carried out based on microscopic data called Transmission Electron Microscopy (TEM) images. The aim of the research is to produce a neural network (CNN/Deep Learning) that has been trained to classify two types of variants of the SARS virus, namely SARS-COV and SARS-COV2. The research phase includes data collection, data pre-processing (consists of the image format conversion and enhancing process), and the classification stage. The classification is carried out using both of the original and enhanced image data. The highest classification accuracy was obtained when the original image data was used, namely 77.66%. It was also found that the classification accuracy increased with an increase in the input image size, but the image data enhancing process used was not able to increase the classification accuracy beyond the classification accuracy achieved when using the original image
Hybrid Fourier Descriptor Naïve Bayes dan CNN pada Klasifikasi Daun Herbal
No full textPlants are vital to human life on earth, and the leaves and their whole parts have many benefits. These parts of the plant can help distinguish between different species. The leaf identification can be performed at any time, while the other parts of the plants can only be identified at a certain time. The study aims to classify two types of herbs i.e. saur-opus androgynous and moringa oleifera, implementing the Fourier Descriptor method to extract the shape and texture features. In the process of classification using the Naïve Bayes method with three types of nuclei (Gaussian, Bernoulli, and Multinomial) and a Convolutional Neural Network. The testing process was carried out using two scenarios, dark and light, where each scenario consisted of 240 images for a total of 480 images divided into 20% of the data testing and 80% of the training data. The Fourier Descriptor-Bernoulli Naive Bayes method gives the lowest accuracy in both light and dark scenarios, at 46% and 52%, respectively. As for the classification of herbal leaves using a combination of the Fourier Descriptor-Convolutional Neural Network method, it is recommended to be used in light image scenarios and Fourier Descriptor-Gaussian Naive Bayes in the dark scenarios because it is able to detect herbal leaf types with 100% accuracy