DirectX doesn't have a concept of scale/rotation/translation, or anything like that. The old fixed-function stuff (no shaders) in D3D9 and prior would use a single transformation matrix for those purposes, but all it did was take the matrix. It was up to the calling code to calculate the required matrix. When you're using shaders everything is handled in your shader code, and D3D doesn't know or care how you handle transformations.

The D3D SDK does come with math libraries that have helper functions and classes for doing this stuff. So there's a function for generation a rotation matrix, or multiplying matrices, or calculating the length of a vector, etc. In fact there's actually two such libraries: D3DX, which is older, and XNAMath, which is newer. XNAMath is probably a bit more awkward to use, because it's designed to take better advantage of SIMD operations (SSE/SSE2 stuff).

Check out the free Khan Academy videos to help catch up on your math. They are very good, and not too dry to follow.

You can probably get by without knowing the linear algebra for a little while, maybe you can get some small game prototypes going. But unfortunately if you don't understand it, you're going to find yourself hitting the same wall over and over again, it will be very frustrating. Pretty much any advanced graphics techniques will require you to be familiar with the concepts of the different transforms, and if you don't have some understanding how it works you won't be able to debug your own code, and it will drive you nuts.