I know that if $L\in REG$ then you can build an automata that accepts $L^*$, but I was wondering if my approach is also good. I thought about showing that $$L^*=\{\epsilon\} \cup \bigcup_{n\in \mathbb{N}}L^n$$ Where $$L^n=\{\omega_1 \ldots \omega_n \colon \omega_1 \in L,\ldots , \omega_n \in L \}$$
And then concluding by induction that $L^n$ is regular for every $n$ (since the set of regular languages is closed under union) and finally that $L^*$ is also regular.
So,
Do I have a mistake here, or is this approach valid? thanks!