Augmented Dynamic Adaptive Model (ADAM) for Predicting Daily Seasonal Data

Utilizing the Augmented Dynamic Adaptive Model (ADAM), researchers take a novel approach to model the BIST 100 index, focusing on coping with daily seasonality. By integrating modeltime's advanced regression capabilities, the approach aims to enhance predictive intervals, crucial for time-sensitive financial markets.

ShareShare

In an innovative application of machine learning, the BIST 100 index has been modeled to build predictive intervals by leveraging the Augmented Dynamic Adaptive Model (ADAM). This approach effectively manages daily seasonality, a crucial aspect for time-sensitive financial markets.

The ADAM method utilizes the modeltime::adam_reg function to predict trends and seasonal patterns without being encumbered by too many exterior regressors. This strategic approach capitalizes on advanced algorithmic capabilities to detect inherent patterns within the data.

Interestingly, the timetk::step_timeseries_signature function, often employed for time series modeling, was avoided due to its limitations in handling excessive exterior factors. Instead, the focus was firmly placed on the algorithm’s strength in harnessing internal data trends and seasonal nuances.

The methodology represents a significant step forward in predictive analytics, offering a refined toolset for economists and market analysts focusing on indices such as BIST 100, where seasonality plays a pivotal role.

By refining the input process and utilizing modeltime's capabilities, analysts can better anticipate market movements, making informed decisions based on robust scientific modeling.

For more insights, read the full article at R-bloggers.

Related Posts

The Essential Weekly Update

Stay informed with curated insights delivered weekly to your inbox.