Another method of assessing a model's performance is by evaluating the model's growth of learning or the model's ability to improve learning (obtain a better score) with additional experience (for example, more rounds of cross-validation).
The information indicating a model's result or score with a data file population can be combined with other scores to show a line or curve, which is known as a model's learning curve.
A learning curve is a graphical representation of the growth of learning (the scores shown in a vertical axis) with practice (the individual data files or rounds shown in the horizontal axis).
This can also be conceptualized as:
The same task repeated in a series
A body of knowledge learned over time
The following figure illustrates a hypothetical learning curve, showing the improved learning of a predictive model using resultant scores by cross-validation round:
Source link...