Background
Stock market process is full of uncertainty; hence stock prices forecasting very important in finance and business. For stockbrokers, understanding trends and supported by prediction ...software for forecasting is very important for decision making. This paper proposes a data science model for stock prices forecasting in Indonesian exchange based on the statistical computing based on R language and Long Short-Term Memory (LSTM).
Findings
The first Covid-19 (Coronavirus disease-19) confirmed case in Indonesia is on 2 March 2020. After that, the composite stock price index has plunged 28% since the start of the year and the share prices of cigarette producers and banks in the midst of the corona pandemic reached their lowest value on March 24, 2020. We use the big data from Bank of Central Asia (BCA) and Bank of Mandiri from Indonesia obtained from Yahoo finance. In our experiments, we visualize the data using data science and predict and simulate the important prices called Open, High, Low and Closing (OHLC) with various parameters.
Conclusions
Based on the experiment, data science is very useful for visualization data and our proposed method using Long Short-Term Memory (LSTM) can be used as predictor in short term data with accuracy 94.57% comes from the short term (1 year) with high epoch in training phase rather than using 3 years training data.
Big data encompasses social networking websites including Twitter as popular micro-blogging social media platform for a political campaign. The explosive Twitter data as a respond of the political ...campaign can be used to predict the Presidential election as has been conducted to predict the political election in several countries such as US, UK, Spain, and French. The authors use tweets from President Candidates of Indonesia (Jokowi and Prabowo), and tweets from relevant hashtags for sentiment analysis gathered from March to July 2018 to predict Indonesian Presidential election result. The authors make an algorithm and method to count important data, top words and train the model and predict the polarity of the sentiment. The experimental result is produced by using R language and show that Jokowi leads the current election prediction. This prediction result is corresponding to four survey institutes in Indonesia that proved our method had produced reliable prediction results.
For specific purpose, vision-based surveillance robot that can be run autonomously and able to acquire images from its dynamic environment is very important, for example, in rescuing disaster victims ...in Indonesia. In this paper, we propose architecture for intelligent surveillance robot that is able to avoid obstacles using 3 ultrasonic distance sensors based on backpropagation neural network and a camera for face recognition. 2.4 GHz transmitter for transmitting video is used by the operator/user to direct the robot to the desired area. Results show the effectiveness of our method and we evaluate the performance of the system.
In this study, we present the first spell error corpus for the Indonesian Language (SPECIL). This corpus provides a comprehensive resource for researchers and practitioners to detect and correct ...spelling errors in Bahasa Indonesia (Indonesian). It should be emphasized that currently, there is no recognized corpus for identifying spelling mistakes in the Indonesian language that has been officially released or made accessible. This study also provides a systematic literature review to identify resources and methodologies for building a corpus for spelling error detection and correction in Indonesia. A corpus was created using a combination of manual and automatic methods. The results of this study are a review of publications relating to corpora and spelling, the novel algorithm of six types of spelling errors, and the production of a corpus comprising over 180,000 tokens in 21,500 sentences, including non-word, real-word, and punctuation errors. Using the developed corpus, various Natural Language Processing (NLP) models, including spell checkers and language models, can be trained and tested to enhance their accuracy and effectiveness in identifying and rectifying errors in Indonesian texts. Moreover, the corpus can be used to develop and evaluate new algorithms and techniques for spelling error detection and correction in Indonesia. The SPECIL corpus is publicly available and accessible. It is expected that SPECIL will inspire further research in this area and facilitate the development of more accurate and effective spelling error detection and correction tools in Indonesian language.
The tourism industry has currently grown in various aspects, including the types of attractions, their quantity, and the number of tourist visits in various regions, contributing positively to both ...regional and global economies. Historical transactions are essential for developing recommender systems, utilizing techniques such as Collaborative Filtering and Demographic Filtering. TripAdvisor is a reputable website providing a wide range of accessible tourism information, including attractions, user profiles, and ratings. However, this unstructured raw data requires processing to create an adequate dataset for recommender systems. This study conducted a series of data processing steps on the raw data, including data restructuring, validation, content addition, integration with Google Maps, normalization, and modeling. This study successfully produced an original dataset comprising User Transaction, Item or Attraction, Attraction Type, Continent, Region, Country, City, and Visiting Mode. It also includes an entity relational model for tourism in Indonesia, particularly in Bali, Malang, and Yogyakarta regions, based on various global user experiences. This dataset is adequate and essential for developing various models of tourism recommender systems such as using Collaborative Filtering.
An essential element of association rules is the strong confidence values that depend on the support value threshold, which determines the optimum number of datasets. The existing method for ...determining the support value threshold is carried out manually by trial and error; the user determines a support value such as 10%, 30%, or 60% according to their instincts. If the support value threshold is inappropriate, it produces useless frequent patterns, overburdens computer resources, and wastes time. The formula for predicting the maximum count of frequent patterns was 2n - 1, where <inline-formula> <tex-math notation="LaTeX">n </tex-math></inline-formula> is the number of distinct items in the dataset. This paper proposes a new SDFP-growth algorithm that does not require manual determination of the support threshold value. The SDFP-growth algorithm will perform dimensionality reduction on the original dataset that will generate level 1 and level 2 smaller datasets, thus automatically producing a dataset with an optimum amount of data with a minimum support value threshold. The proposed formula for predicting the maximum number of frequent patterns will become 2<inline-formula> <tex-math notation="LaTeX">^{\vert A\vert } </tex-math></inline-formula> - 1, which is <inline-formula> <tex-math notation="LaTeX">\vert A \vert </tex-math></inline-formula> will always be smaller than <inline-formula> <tex-math notation="LaTeX">n </tex-math></inline-formula>. Experiments were performed on five various datasets, which reduced the number of data dimensions by more than 3% on the Level 1 dataset and more than 69% on the Level 2 dataset by maintaining the confidence value of the strong rules. In the execution time evaluated, we found an optimization of more than 2% on the level 1 dataset and more than 94% on the level 2 dataset.
The recommender system has gained research attention from education research communities mainly due to two main reasons: increasing needs for personalized learning and big data availability in the ...education sector. This paper presents a hybrid user-collaborative, rule-based filtering recommendation system for education context. User profiles are represented by learning outcome scores and contextual information. The user-collaborative filtering method is used for predicting the targeted student’s learning outcome of a particular course. The predicted learning outcome combined with a set of decision rules are used for recommending some relevant link of learning materials to the targeted student. The initial contextual information which is assessed during the online program entrance test makes it possible for the proposed recommneder system to give automated recommendations to new students. The proposed method was tested using student learning outcome records from BINUS Online repository data. The results of performance evaluation of both recommender system with contextual information which achieves 458.22 MSE and the recommender system without contextual information which achieves 413.19 MSE are not significantly different. However, unlike the latter recommender system, the former recommender system has an advantage mainly that it can be used to give recommendation to the targeted students since their early program stage.
•ATS and SA are essential for customer feedback analysis.•Airline reviews contain intricate narratives and multiple aspects.•Domain shift occurs when the target domain exhibits out-of-distribution ...data.•DA via two-stage finetuning addresses the domain shift issue for ABS.•Higher customer ratings are linked to positive recommendations and positive sentiment valence.
Opinion summarization and sentiment classification are key processes for understanding, analyzing, and leveraging information from customer opinions. The rapid and ceaseless increase in big data of reviews on e-commerce platforms, social media, or review portals becomes a stimulus for the automation of these processes. In recent years, deep transfer learning has opted to solve many challenging tasks in Natural Language Processing (NLP) relieving the hassles of exhaustive training and the requirement of extensive labelled datasets. In this work, we propose frameworks for Abstractive Summarization (ABS) and Sentiment Analysis (SA) of airline reviews using Pretrained Language Models (PLM). The abstractive summarization model goes through two finetuning stages, the first one, for domain adaptation and the second one, for final task learning. Several studies in the literature empirically demonstrate that review rating has a positive correlation with sentiment valence. For the sentiment classification framework, we used the rating value as a signal to determine the review sentiment, and the model is built on top of BERT (Bidirectional Encoder Representations from Transformers) architecture. We evaluated our models comprehensively with multiple metrics. Our results indicate competitive performance of the models in terms of most of the evaluation metrics.
Background
The main obstacle for local and daily or weekly time-series mapping using very high-resolution satellite imagery is the high price and availability of data. These constraints are currently ...obtaining solutions in line with the development of improved UAV drone technology with a wider range and imaging sensors that can be used.
Findings
Research conducted using Inspire 2 quadcopter drones with RGB cameras, developing 3D models using photogrammetric and situation mapping uses geographic information systems. The drone used has advantages in a wider range of areas with adequate power support. The drone is also supported by a high-quality camera with dreadlocks for image stability, so it is suitable for use in mapping activities.
Conclusions
Using Google earth data at two separate locations as a benchmark for the accuracy of measurement of the area at three variations of flying height in taking pictures, the results obtained were 98.53% (98.68%), 95.2% (96.1%), and 94.4% (94.7%) for each altitude of 40, 80, and 100 m. The next research is to assess the results of the area for more objects from the land cover as well as for the more varied polygon area so that the reliability of the method can be used in general
Automated question generation is a task to generate questions from structured or unstructured data. The increasing popularity of online learning in recent years has given momentum to automated ...question generation in education field for facilitating learning process, learning material retrieval, and computer-based testing. This paper report on the development of question generation framework based on key-phrase method for online learning with a constraint that the generated questions should comply with the learning outcomes and skills from Bloom’s Taxonomy. The proposed method was tested using learning materials of Software Engineering course for undergraduate level written in Bahasa Indonesia obtained from Bina Nusantara’s (Binus’s) Online Learning repository. Using one-semester lecture material, this study generated 92,608 essay-type questions from 6-level Bloom’s Taxonomy which were further sampled randomly to obtain 120 question samples for method evaluation. Performance evaluation using average Bilingual Evaluation Understudy (BLEU) involving five independent reviewers toward samples of these questions achieved 0.921 and 0.6 Cohen’s Kappa. The relevance of Bloom’s Taxonomy level of the generated questions was evaluated by means of classification model with 0.99 accuracy. The results indicate that not only are the generated questions well understood and agreed by the reviewers, they are also relevant to the expected Bloom’s Taxonomy level there for the questions can be delivered to students in the respected course delivery and evaluation.
PDF
