Super early prototype as you can tell from the fact that the houses are gobbledegook!

The idea should be pretty self explanatory. First enter the County you'd like to buy in and the size of deposit you can afford. Then move the slider to give the size of the mortgage payments you can afford each month. Hit submit and the site gets back to you with a list of properties in your price range.

The rates it uses in the background to calculate the house price you can afford are nonsense for the moment but eventually I'd hope to hook it up to real rates. Shouldn't be hard assuming I can actually get access to information like that.

Thoughts?

term of mortgage?

It's defaulting to 25.

Obviously it's easy enough to add another input to take the term but I'd like to keep the number of inputs the people have to fill to a minimum if possible so I was thinking about making the results tabbed so you could flick between 25, 30 and 35 year terms.

Would that be satisfactory or should I just add the input?

Great idea, fair play to you.

Put in the options for mortgage length etc., but I would have "Click here for recommended settings" button that sets it to 25 years at 7.5%

How do you get the housing data, do myhome or daft provide web services?
What are you using on the server-side?

Good idea, but if it becomes popular I expect myhome and daft will either copy it or shut of your access to their data.

Number of bedrooms would be a fairly important variable, i would have thought?

Fantastic idea though.

Daft have a search API but it's limited to estate agents only for the foreseeable future so that's a no go. MyHome doesn't expose any of it's services afaik. I guess I'd have to actually approach estate agents myself to build a catalogue. I have the bulk of the code needed for adding houses in place already.

The more difficult part I'd guess is getting my hands on reliable mortgage rate info for different deposit percentages. I've no idea if banks can provide these services and if they would let me have access to it. I definitely definitely do not want to force people to choose their own interest rate though because it 1) forces users to know something about current interest rates and 2) just gives people the option to fool themselves.

Server side it's just a barebones Rails app with Search and House models. Super basic. Not sure if that's what you were asking?

And yes, I'd imagine Daft could crush me if they wanted to. Can't think of any good way to exclusivise the service.

Yep. I'm thinking along the lines of a bunch of slide out advanced search controls which are optional but can help narrow the search down. More accurate location, beds etc. All the usual stuff. Also I'd like to let people narrow down the mortgages so that, for example, it only used mortgage rates from 2 year fixed rate mortgages by BOI.

I built some webscraping tools for my price/sq.ft. analysis. It scrapes price, house type, sale/auction, number of beds, square footage if available, url link, from myhome.ie listings. Only hits their summary search pages, so scrapes their entire catalogue reasonable quickly. Written in Java -- you are welcome to the source code if it is of any use to you. It's prone to breakage when they change their listing formats, but you just have to tweak some pattern matching regexes. Web scraping, pattern matching, and analysis portions are three separate programs, so you can pick and choose what you want to re-use.

Alternatively, I can provide you the post-scraping-and-matching database in CSV or other format (e.g. XML, H2 relational database), containing the above fields from the myhome.ie catalog. Latest scrape is as of 1-Apr-2011. I plan to run them quarterly for price analysis.

I like, and would definitely use.

Hi.
Thanks for the offer, I'd love to get a look at that for sure. What's the best way to get it to me? Can you put it on github?

Why'd you choose to scrape myHome as opposed to Daft?

Actually you know what I think I'm ok. I just banged some ruby together to scrape daft. It seems pretty easy to do and it works well. Thanks anyway.

Also I meant to say that I'll give you my scraper code when I'm finished if you like?

Sorry for the delay. I was going back to, uh, comment the code a bit.
Sounds like you are sorted now anyway. In response to your earlier question about myhome vs. daft, myhome started carrying square footage in a structured (i.e. scrapable) form.

Anyway, here's the "spec" for my stuff, in case you want to compare notes:

1) Download all of "http://www.myhome.ie/residential/%s/property-for-sale?page=%s",
where the first placeholder is from the list of 56 counties/postcode/areas on myhome, and the second is a number that you increment until no results are returned.

2) Massage the downloads to be valid xhtml (the myhome site has some warts in this regard)

3) Use an XPath expression to locate the results within each page.

4) Use a regex to parse each result into price, number of beds, area (if present), etc. etc.

I dumped the output in CSV format, and loaded it into the free H2 relational database for analyses.