I conducted a survey asking people of various income groups about their annual expenditure under various consumption brackets such as food, clothing, housing etc. I then calculated the amount spent by people of different income brackets in these categories as a percentage of their total income. Now, to plot this data and find the best fit line which type of regression analysis should I use. Also are there some other good statistical tools that I can use for exploiting this data?
Your charts will depend on what story you are interested in. Are you interested in amounts spent (on average the rich usually spend more on most items than the poor do)? Or are you interested in the shares of expenditure (or income) going on different items? Can you avoid your charts getting too cluttered? Do you intend to adjust your numbers for household size? Or household type, especially whether there are children?
You might get some ideas from the UK publication Living Costs and Food Survey. You will find that linear regression of expenditure on income is often not particularly appropriate.
I don't think regression will serve your purpose. You seem to want to show how expenditure breakdowns differ among people of different income categories. Imagine you have a table with, say, 5 rows representing the income groups and 6 columns representing types of expenditures. Within each cell would be a dollar amount. You'd want to show where there are disproportionalities from row to row. Do you see what statistical test would be applicable there?