Forecasting of Milk Production in Northern Thailand Using Seasonal Autoregressive Integrated Moving Average, Error Trend Seasonality, and Hybrid Models

Milk production in Thailand has increased rapidly, though excess milk supply is one of the major concerns. Forecasting can reveal the important information that can support authorities and stakeholders to establish a plan to compromise the oversupply of milk. The aim of this study was to forecast milk production in the northern region of Thailand using time-series forecast methods. A single-technique model, including seasonal autoregressive integrated moving average (SARIMA) and error trend seasonality (ETS), and a hybrid model of SARIMA-ETS were applied to milk production data to develop forecast models. The performance of the models developed was compared using several error matrices. Results showed that milk production was forecasted to raise by 3.2 to 3.6% annually. The SARIMA-ETS hybrid model had the highest forecast performances compared with other models, and the ETS outperformed the SARIMA in predictive ability. Furthermore, the forecast models highlighted a continuously increasing trend with evidence of a seasonal fluctuation for future milk production. The results from this study emphasizes the need for an effective plan and strategy to manage milk production to alleviate a possible oversupply. Policymakers and stakeholders can use our forecasts to develop short- and long-term strategies for managing milk production.

Learning Future-Aware Correlation Filters for Efficient UAV Tracking

In recent years, discriminative correlation filter (DCF)-based trackers have made considerable progress and drawn widespread attention in the unmanned aerial vehicle (UAV) tracking community. Most existing trackers collect historical information, e.g., training samples, previous filters, and response maps, to promote their discrimination and robustness. Under UAV-specific tracking challenges, e.g., fast motion and view change, variations of both the target and its environment in the new frame are unpredictable. Interfered by future unknown environments, trackers that trained with historical information may be confused by the new context, resulting in tracking failure. In this paper, we propose a novel future-aware correlation filter tracker, i.e., FACF. The proposed method aims at effectively utilizing context information in the new frame for better discriminative and robust abilities, which consists of two stages: future state awareness and future context awareness. In the former stage, an effective time series forecast method is employed to reason a coarse position of the target, which is the reference for obtaining a context patch in the new frame. In the latter stage, we firstly obtain the single context patch with an efficient target-aware method. Then, we train a filter with the future context information in order to perform robust tracking. Extensive experimental results obtained from three UAV benchmarks, i.e., UAV123_10fps, DTB70, and UAVTrack112, demonstrate the effectiveness and robustness of the proposed tracker. Our tracker has comparable performance with other state-of-the-art trackers while running at ∼49 FPS on a single CPU.

The effect of the TF-IDF algorithm in times series in forecasting word on social media

Forecasting is one of the main topics in data mining or machine learning in which forecasting, a group of data used, has a label class or target. Thus, many algorithms for solving forecasting problems are categorized as supervised learning with the aim of conducting training. In this case, the things that were supervised were the label or target data playing a role as a 'supervisor' who supervise the training process in achieving a certain level of accuracy or precision. Time series is a method that is generally used to forecast based on time and can forecast words in social media. In this study had conducted the word forecasting on twitter with 1734 tweets which were interpreted as weighted documents using the TF-IDF algorithm with a frequency that often comes out in tweets so the TF-IDF value is getting smaller and vice versa. After getting the word weight value of the tweets, a time series forecast was performed with the test data of 1734 tweets that the results referred to 1203 categories of Slack words and 531 verb tweets as training data resulting in good accuracy. The division of word forecasting was classified into two groups i.e. inactive users and active users. The results obtained were processed with a MAPE calculation process of 50% for inactive users and 0.1980198% for active users.

Development of AI-Based Diagnostic Model for the Prediction of Hydrate in Gas Pipeline

For the stable supply of oil and gas resources, industry is pushing for various attempts and technology development to produce not only existing land fields but also deep-sea, where production is difficult. The development of flow assurance technology is necessary because hydrate is aggregated in the pipeline and prevent stable production. This study established a system that enables hydrate diagnosis in the gas pipeline from a flow assurance perspective. Learning data were generated using an OLGA simulator, and temperature, pressure, and hydrate volume at each time step were generated. Stacked auto-encoder (SAE) was used as the AI model after analyzing training loss. Hyper-parameter matching and structure optimization were carried out using the greedy layer-wise technique. Through time-series forecast, we determined that AI diagnostic model enables depiction of the growth of hydrate volume. In addition, the average R-square for the maximum hydrate volume was 97%, and that for the formation location was calculated as 99%. This study confirmed that machine learning could be applied to the flow assurance area of gas pipelines and it can predict hydrate formation in real time.