In the plot that looks like a line (21a), it looks like the observations in the data have been sorted from lowest to highest value of the response, yTrainN.
In the plot that looks "filled in" (21b), it looks like the values of the response, yTrainN, appear in a random order in the data.
For a more detailed response, please share your data and/or a more detailed view of your workflow. Some questions to consider:
(1) Before loading the data into Regression Learner, was the data sorted differently in R2021a and R2021b?
(2) In the session start dialogue in Regression Learner, were different options selected in R2021a and R2021b?
