I'm relatively new to machine learning and just trying to have a solid understanding of the basics.
I scraped a real estate/house prices dataset off a website, which I am in the process of cleaning before modelling.
I noticed that out of 1,998 examples/rows in my dataset, 19 of them have a missing target variable (House Price). (I have 11 features in this dataset)
There is such a thing as imputing the target variable, but I believe this is more for quantifying categorical targets for modelling.
So my question is: should I impute the missing target price values with , say, the average price of the rest of the dataset? Or should I just drop the 19 rows with the missing price target values?
Since the whole point of this exercise is to train a model to predict the target with this training data, and these training examples have a missing target, then I cannot use them. So my gut instinct is to drop them from the training set (and maybe later use them as a test set). If I were to impute the 19 target price values and then use the rows in my training set, I'm guessing the accuracy of the model would be affected somehow.
Please help me understand what is the correct line of thinking here.