As you can see from the above image, confident learning is about estimating the likelyhood of the data being labeled correctly based upon the confidence of the model. If the model confidence is above the threshold confidence (The Tj parameter, tdog, tfox tcow) and if the confidence of the model prediction is higher than the threshold but the label is different, then we predict a wrong label
Data drift refers to the phenomenon where the statistical properties of a dataset used for machine learning or analysis change over time. This alteration can be due to various factors, such as shifts in data collection processes, changes in the underlying distribution of the data, or modifications in the environment from which the data originates. Detecting and addressing data drift is crucial to maintaining the performance and reliability of machine learning models and analytical systems.