Kaiho Cheung¹, Ishmael Rico², Tao Li³ and Yu Sun⁴, ¹Sentinel Secondary, Canada, ²University of California, USA, ³Purdue University, USA, ⁴California State Polytechnic University, USA

In recent years the popularity of anime has steadily grown. Similar to other forms of media consumers often face a pressing issue: “What do I watch next?”. In this study, we thoroughly examined the current method of solving this issue and determined that the learning curve to effectively utilize the current solution is too high. We developed a program to ensure easier answers to the issue. The program uses a Python-based machine learning algorithm from ScikitLearn and data from My Animelist to create an accurate model that delivers what consumers want, good recommendations [9]. We also carried out different experiments with several iterations to study the difference in accuracy when applying different factors. Through these tests, we have successfully created a reliable Support vector machine model with 57% accuracy in recommending users what to watch.

Machine learning, anime, recommendations, Python.