The neural network was written in JAVA without any use of machine learning libraries. Feed forward, error calculation, back propagation, weight updates, and all necessary matrix operations were put down from scratch. You can find the source code published under GPL here.
TSMLTC was developed with Angular (Type Script), Spring Boot (Java), Hibernate/JPA (object-relational mapping), PostgreSQL (relational database) and Apache Kafka (Messaging). Access to neural network and regression analysis is accomplished via REST interface and Apache Kafka. Time series, regression values and network forecasts are visualized graphically with plotly (Java Script).