UPLB ICS Peak One

Small and Medium Enterprises (SMEs) would employ the use of Enterprise Resource Planning (ERP) Systems to automate processes in certain facets of a business. For SMEs to slowly integrate their ERP systems without affecting day to day operations, the systems are split into modules that would only implement a portion of the business operation, like sales, collections, and inventory. To expand decision making tools and to facilitate increasing orders, a sales, collections, and finished goods inventory logging system is developed for the single proprietor business Jam’s Bakery. The standalone application is created using JavaFX with MySQL covering the database. The usability of the application is evaluated using the System Usability Scale (SUS). The application performed well in system usability, but there’s a need for improvements in clarity of objectives and simplification of workflow.

The UPLB TAKAM mobile application represents a timely and essential solution post pandemic, addressing the challenges that UPLB students face in meal preparation, budgeting, and nutrition. Through its innovative features such as intelligent ingredient management, personalized meal plans, and budget-friendly recipes, UPLB TAKAM aims to alleviate the burdens of time constraints, financial limitations, and unhealthy eating habits. By providing convenient access to budget-friendly, nutritious meals, UPLB TAKAM helps students make healthier food choices, develop cooking skills, and improve their well-being, supporting long-term health in a changing educational and global environment.

Air pollution is one of the most serious environ- mental problems nowadays. Forecasting the air quality index (AQI) level is beneficial for preventing further harm and creating effective solutions. This study attempted to forecast the air quality index level using machine learning. Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) were used to build the models. Selected historical data from the World Air Quality Historical Database Website were processed for training the models. The performance of the models was evaluated using accuracy and f1-score. For the training data set, the LSTM model obtained an accuracy of 88.86% and an f1- score of 88.93%, while the CNN model obtained an accuracy of 88.04% and an f1-score of 88.23%. For the testing data set, the LSTM model obtained an accuracy of 83.93% and an f1- score of 83.59%, while the CNN model obtained an accuracy of 82.90% and an f1-score of 82.70%. However, when predicting the minority class, the CNN model obtained an accuracy of 20% and a f1-score of 31.57%, whereas LSTM performed worse, having an accuracy of 4.44% and a f1-score of 8.33%. It can be concluded that, while the LSTM model has better general performance, the CNN model was more effective in handling underrepresented classes.

AI has changed how higher education and academic institutions connect with their students. The study aims to create a chatbot for college admissions inquiry system. A chatbot that is capable of handling the queries of the applicants can save time and effort for both the applicant and the staff member. The chatbot aims to help the applicants by providing the answers and information they need. Further, it seeks to aid the institution by giving assistance and support to the staff in responding to the applicants. The chatbot produced an accuracy rate of 98.84%. Also, it achieved 99.42% and 98.84% as precision and f1-score respectively. The chatbot application obtained a mean CUQ score of 80.1 (SD 9.1). It indicates that the chatbot’s personality, onboarding, user experience, and error management are above average. Feedback highlighted the chatbot’s informativeness, reliability, and user-friendliness. Overall, this study validates the potential of chatbots in enhancing the admissions inquiry process, providing a scalable and cost-effective solution for educational institutions.

Dogs have a unique role in our modern society. They have been a part of a lot of families, assisting humans in their day-to-day lives. Just like humans, dogs are also capable of feeling different kinds of emotions which they can show through their body language [1]. Recognition of these emotions from dogs will allow humans to make better decisions around them. A dog’s visual expressions can be extracted into features, which makes a Convolutional Neural Network powerful when it comes to classifying dog emotions based on images. In this proposal, a fine-tuned model based on the MobileNet architecture was trained over 52517 images of dogs containing angry, happy, relaxed, and sad emotions. Based on the results, the model will use the fold with an accuracy of 0.8418 and an average F1 score of 0.838. A Flutter-based mobile application was developed for the model to be accessed even without an internet connection. The application was tested manually with a set procedure. A System Usability Scale was used to assess the usability score of the system, garnering a score of 79.83, which indicates good usability.