The student can calculate model predictions using predict() , knowing that this function takes a model object and a data frame as arguments.

The student knows that Python classes are initialized with a method named __init__()  and that this method takes self  as the first argument.

The student can calculate groupwise summary statistics with dplyr , by first calling group_by()  then calling summarize() .

The student can articulate these modeling workflow steps: Split the data into training and testing sets, then train the model on the training set, then test the predictions on the testing set.

The student understands the definition of autocorrelation.

The student can predict what happens when an error occurs inside a try-block.

The student can calculate how depreciation affects the value of a thing over time.

The student can interpret that on average R programmers live eight years longer than SAS programmers.

The student can interpret that Python programmers are ten times more likely to own a pet snake than non-Python programmers.

Having too many learning objectives. In general, an exercise should teach the student precisely one thing. If you want to have multiple learning objectives, it's a warning sign that you are trying to cram too much into a single exercise. That said, if you have several simple learning objectives, a multi-step exercise can be appropriate.

The exercise doesn't test the learning objectives. It is important that the exercise meets the objective. One common danger is that you start writing code and somehow the exercise ends up being completely different from what you originally intended. This can be fine, but make sure that you are teaching the things that students will need to know for future exercises.