Draconis spelled out some of the problems in trying to decide what the specific rules are, given a lack of data. I'd like to address the question of whether there are rules at all. The teacher seems to have given a model of such a "rule", where *pinta → winta 'arm'. The use of the star tells me that this is historical change, this is not about phonological rules; the reconstructed form is *pinta, and some set of sound changes gave winta. The question is, what are those changes? If the exercise is about historical change, the rules of the game are very different from synchronic phonological analysis. A possible solution to a p-to-w change is p→pʰ→φ→ʍ→w.
In Javanese, there is a contrast between dental and retroflex, the latter notated with an added h. If we assume that the reconstructed root is *andek, there are four questions to be answered: what is the cause of the initial consonant, how did the stop become retroflex, why did the first vowel become schwa, and why did e become [ɛ] in a closed syllable (this is an actual phonological of Javanese)? We also have to know what dialect of Javanese this is.
There is a somewhat-open question as to whether proto-Austronesian has retroflex as well as alveolar consonants: therefore the assumption that the historical input has an alveolar cannot be made gratuitously.
Blust claims that it is common in Austronesian for t to be postdental and for d to be alveolar (he objects to calling the back linguals "retroflex"), where Javanese innovated the contrast by filling in voiced postdentals and voiceless alveolars (due to large-scale borrowing). The presence of a retroflex consonant in Javanese is often taken to be evidence of loanword status. The question "how did d→ɖ?" assumes, without evidence, that this is a sound change. But if this is a loan word, then the correlation has a different interpretation – ɖ is the ordinary re-interpretation of a "foreign" lingual voiced stop, upon borrowing into Javanese.
Blust's book is an excellent starting for figuring our how these changes might have come about. But before trying to figure out how something happened, you need evidence that something did happen: reconstructions, like synchronic underlying forms, are not self-evident or God-given.