The single most useful classification of viruses is by genome. The Baltimore scheme groups viruses by how their genome is converted into messenger RNA, because every virus must produce mRNA that the host ribosome can read. The host ribosome only reads single-stranded RNA in the positive sense — the same sense as mRNA. Anything else has to be converted first, and the conversion step determines where replication happens and what the virus must carry with it.
A positive-sense single-stranded RNA genome can be translated directly by ribosomes the moment it enters the cytoplasm. No packaged polymerase is strictly required, because the first protein made can be the polymerase itself. A negative-sense RNA genome is the complement of mRNA, so it cannot be translated; the virus must package an RNA-dependent RNA polymerase to copy it into positive-sense mRNA. Double-stranded RNA genomes face the same problem and also package a polymerase, typically inside the capsid.
DNA viruses generally replicate in the nucleus, where host DNA and RNA polymerases and the nucleotide pool are available. Most DNA viruses use host polymerases for at least part of their transcription, though large DNA viruses often encode some of their own replication enzymes. Reverse-transcribing viruses carry the machinery to copy RNA into DNA and then integrate that DNA into the host chromosome, which requires nuclear access.
Segmented genomes — genomes split into separate nucleic acid molecules — allow reassortment, the swapping of whole segments between related viruses. That is a distinct mechanism from mutation and becomes important later when we discuss antigenic shift.
A practical rule: if the genome cannot be read directly by a ribosome, the virus must bring or immediately build the enzyme that makes it readable, and that constraint shapes where in the cell the whole process unfolds.