EasyAI

EasyAI is a module that was developed for those who have little or no knowledge about artificial intelligence models, but want to achieve good results.

Based on your data, this module will test various AI models suitable for your dataset and will prepare the one that achieved the best result for use. The necessary information will be saved in a file, which can be distributed and used later on any terminal for predictions on various data.

The module uses cross-validation, which means it performs 4 rounds of training with each possible model. In each round, ¼ of the data is separated and does not participate in the training. These separated data are used to test the model, one by one. At the end of the process, all the data from your dataset will have been tested by a model that has never seen them before. The best model is then saved in the file.

Currently, it can be used to perform 3 tasks:

• Regression: A model that predicts numerical values, such as the price of a house or tomorrow's temperature.

• Classification: A model that categorizes data into groups, such as identifying whether an email is spam or not.

• Recommendation: A system that suggests items or content based on previous preferences or behavior, such as movies or products.

  • Recommendation for item: Recommends similar items to the requested item.

  • Recommendation for user: Recommends items that similar users have consumed most and highly rated but the requested user has not.

It is crucial that the data makes sense with the outcome, as artificial intelligence models learn by analyzing patterns in the data. If the data is not relevant or consistent with what is being predicted or classified, the model will not be able to identify the correct relationships and will produce inaccurate or useless results. Bad data leads to bad predictions.