School of Engineering Department of Computer Science and Engineering 101 AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN Gary Shueng Han / CSE Student: CANDICE Angeline / MATH-CS Course: UROP1000, Summer This report features a localization technology based on the SiFu framework by feeding spot-based and sequence-based signals into a likelihood fusion and employing a particle filter to map indoor location. In addition, it studies the particle filtering in more detail and presents the mathematical formula used to estimate the distribution using particle filter. After a brief introduction to the fusion framework, this paper shifts its focus to debugging a fusion localization engine on a cloud environment. By changing a Singleton class comprising of static pointers into separate objects for each machine, in order to be less computationally intensive and hence more energy efficient. Lastly, this report suggests a real-world application of localization technology in aiding wandering elderlies. AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN Gary Shueng Han / CSE Student: WIDJAJA Oscar / COGBM Course: UROP1000, Summer Trajectory recovery is the practice of extracting a path from discrete location data, which are present in massive amounts due to advances in sensors technology and Internet of Things. Nevertheless, this task is often difficult since data is usually noisy, sparse, and sporadic. This project implemented 3 models for trajectory recovery: simple linear implementation, and two Random Forest models with different sets of features. Data was first pre-processed with NumPy into Trajectory and Point objects. They were then laid out in a grid and different models using scikit-learn were used to predict simulated missing location data. However, model performance gave unexpected results such as increases in accuracy when data is sparser. Therefore, future work is needed to revisit the implementation for more meaningful result. AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN Gary Shueng Han / CSE Student: AGRAWAL Himalaya / COMP Course: UROP1100, Summer Dyslexia is a learning disorder which makes it difficult to read and interpret words and letters but does not affect general intelligence. In this project, our goal was to help pre-screen dyslexic students so that we could provide them with the extra attention and assistance they may require in order to read and learn effectively. This was to be done by using the data collected in a Chinese writing application developed for dyslexia prescreening, where the students would write the characters as instructed, in the hopes of finding any relationships between the data and dyslexia in particular students.