1. General info
Dodona in Epirus in northwestern Greece was the oldest Hellenic oracle, possibly dating to the second millennium BCE according to Herodotus. The earliest accounts in Homer describe Dodona as an oracle of Zeus. Situated in a remote region away from the main Greek poleis, it was considered second only to the Oracle of Delphi in prestige. Priestesses and priests interpreted the rustling of the leaves of the oak trees by the wind to determine the correct actions to be taken.
Looking back at the history of the game of predicting, one can conclude that it remained the same. The art of prediction has always been an extremly valuable skill, and the ability to anticipate future events and trends can be incredibly lucrative in many fields.
While the tools and methods used for prediction have evolved over time, the underlying principles have remained the same. Successful predictors are able to gather and analyze data, identify patterns and trends, and use their insights to make informed decisions.
Nevertheless, predicting the future is not an exact science. Even the most skilled and experienced predictors can be wrong, and unforeseen events can throw carefully crafted predictions off course. That being said, those who are able to consistently make accurate predictions are highly valued in many industries.
2. Predicting electricity price
Electricity price prediction facilitates efficient energy management, enables informed decision-making, reduces costs for consumers, and supports the stability and optimization of energy markets and grid operations.
Electricity price prediction benefits various stakeholders:
- Producers: Optimize production and scheduling, maximize revenue.
- Consumers: Manage energy usage, budget effectively.
- Traders: Identify trading opportunities, mitigate risks.
- Grid Operators: Plan for demand, maintain grid stability.
- Regulators and Policy Makers: Evaluate market performance, inform regulations.
Here we plot two predictions, for day-ahead, and prediction from the previous day with a given real data. The predictions are for the day ahead auction prices with delivery on the Serbian control area. Beside this we draw the region of confidence of 90% calculated with conformal prediction
4. Conformal prediction
Conformal prediction is a framework that provides valid and reliable prediction intervals for machine learning models. When specifying a conformal prediction with an area of 90%, it means that the prediction interval will contain the true value with a probability of 90%.
In simpler terms, if a conformal prediction model with an area of 90% is applied, it will generate prediction intervals that are expected to capture the actual value 90% of the time. This level of confidence can be useful in various applications where uncertainty estimation is critical, such as financial forecasting, medical diagnosis, or environmental monitoring.