Many of the same algorithms that we've seen so far can also be used for numeric targets, where you are trying to predict something like a dollar amount or another continuous value. There is an easy way to keep track of which algorithms are used for categorical targets only, numeric targets only, or both. If you look at the Expert tab in an Auto node like the Auto Numeric node you can see a list of the algorithms that can be used to make the kind of prediction shown in the following screenshot. We will discuss all three Auto nodes in more detail in the next section:

Notice that CHAID and Neural Net, among others, appear in this list and also appeared listed in the previous section. They can be used with either categorical or continuous targets. Notice also that Regression and Linear appear in this list, but they did not appear in the previous list. This is because they can only be used to perform numeric target predictions. So the best way to remember is to check both in this node and the Auto Classifier node to see what is available. It is recommended to learn at least one categorical classifier and one numeric classifier before you start to use the Auto nodes.

Linear Regression is the most famous of these algorithms. If you are already familiar with regression, then the Regression node will be closer to what you are expecting. It is the plain vanilla regression you would encounter in a university setting. The Linear node has some impressive features, but does a lot behind the scenes. It will discard weak variables and perform other kinds of automatic data preparation for you. It might be a good idea to develop a familiarity with the advanced features of the Linear node before you trust it to run without oversight. Even when its settings are consistent with your goals, you will want to understand what it is doing.