Chronolang: A Domain-Specific Language for time series analysis and forecasting

Abstract

This study introduces ChronoLang, a domain-specific programming language (DSL) designed for time series analysis and forecasting, which aims to address the limitations of existing tools when working with historical and real-time data processing. Along with the analysis of domain requirements, grammar design, and the implementation and development of a modular compiler architecture, the methodology involved a critical evaluation of existing tools like Pandas, SQL, and Apache Spark. In addition to its built-in forecasting capabilities using ARIMA, Prophet, and LSTM models, ChronoLang provides a native declarative model for temporal queries and extensible integration with a variety of data sources, such as CSV files, SQL databases, and MQTT/Kafka streams. By combining native streaming support and predictive analytics into a single, unified environment, ChronoLang tries to eliminate most of the architectural fragmentation found in conventional solutions. The preliminary results show significant improvements in workflow simplicity and execution efficiency. The study highlights ChronoLang's potential to simplify analytical workflows in multiple domains, including climate research, industrial IoT, and finance.