You have some examples available in the documentation.
Basically you need to build your own analyzer with a edge ngram token filter and use this custom analyzer at index time (in the mapping as the analyzer for your field). I'd recommend setting a search_analyzer in the mapping as well for the same field and set it to simple.
I'd also recommend that you use the _analyze API to understand what elasticsearch is going to do at index time and at search time.
I'd not use Analyzer here.