Linear Regressions are regression models. They are very simple, because they assume that the fields in your Dataset will be related to the field being predicted in a linear way. When that assumption is not right, they can still be built but their predictions will be wrong. They cannot be used in classification problems and are less informative than Decision Trees.
To learn about all their available configuration options using BigML's Dashboard, you can check the corresponding section of the Linear Regressions documentation. Also the Linear Regressions API documentation contains the information about the attributes that need to be specified to configure your source programmatically.