Supervised learning is where known input data is linked to known output data, so that the alrgorithm can use training data to determine the correlation, with the expectation that it will then be able to take input data that was not part of the training data set and come up with the correct output. Qualitative data is referred to 'classification,' as quantitative is to 'regression.'

Conversely, unsupervised learning is where only the input data is provided, with the expectation that the algorithm will find patterns in the inherent distinctiveness of the data. The objective of this is to establish a model which may include similarities ('clustering') or relatedness ('association') among quantitative qualities.

Semi-supervised learning is neither purely one or the other, whereby only some of the output data is available. This type facilitates advantages of both other types. Data that is not labeled is much more accessible and ubiquitous.

XGBoost (eXtreme Gradient Boosting) is open-source ML algorithm software used by some businesses and in ML competitions. It focuses on gradient boosting, which has exploitative access to many different models and decision trees to arrive at better results. xgboost.ai

---