Time series forecasting is the science of predicting future events based on historical data, a practice with applications that permeate our daily lives. Consider demand and inventory planning, where forecasting enables businesses to anticipate customer needs, ensuring optimal product availability while minimizing costs.

First, we explore the core concepts of foundation models, such as pretraining, transfer learning and fine-tuning. Second, we take a look at the advantages and disadvantages of foundation models in time series forecasting. While they can speed up the modeling and inference process, they might also not be the best solution for a particular project, meaning that we must still have a certain expertise to use them correctly and compare them with other methods. Then, we explore some of the major contributions to the field, including TimeGPT, Chronos, Moirai and TimesFM. We quickly discover their architectures, their capabilities and their limitations. Finally, we see TimeGPT in action to demonstrate how a foundation model can be used and how it compares to traditional methods.

In this talk, we introduce a novel approach leveraging genetic algorithms to optimize both forecast stability and accuracy, creating a dynamically weighted ensemble that balances these competing objectives and delivering better accuracy than any single base model. By incorporating past model performance into our evolutionary framework, we iteratively evolve an ensemble that minimizes large forecast swings while maintaining or improving overall accuracy. We demonstrate how this method systematically adjusts model weights based on historical deviations and performance metrics, solving a key business challenge.

Join dotData Sr. Data Scientist Sharada Narayanan as she dives into the strengths, uses, and limitations of popular time-series forecasting techniques like ARIMA and Prophet.

Sharada will walk through real-world examples, share code snippets, and explore how ARIMA Prophet compare when building models using Feature Engineering techniques and advanced machine learning algorithms. Dive into the insights of each method and see how programmatic feature engineering and machine learning can supercharge your time series analysis.

Dr. Yasin Ceran presents a comprehensive workshop on time series forecasting models and techniques.