[Dave] Rich, you better do this one, because I'm still calming down *Laughs*
[Rich] What are you calming down from Dave?
[Dave] Well I just had my session, and it seemed to go quite well and I'm here with the top speaker so far, Rafal
[Rich] Well I'm here with the Top speaker and the 7th top speaker at Tech Ed. So I'm, you know, I'm in real honoured company
And I'm here with the top 10 speakers
[Dave] Fantastic, Rafal, how are you?
I'm very well thanks you, Oh I'm tired but I'm so happy
[Dave] Yes, it's been a long old week, and I believe you got another week of it yet?
Yeah, I'm excited, I'm elated, now I'm going to have this wonderful dinner they'll take me out to tonight
[Dave] Yeah it's a nice nice place. So what have you been speaking on this week Rafal? You had a whole bunch of sessions haven't you? What have you been doing?
Erm, I guess I've been covering the usual suspects. So I've had a bit of security, I had a kinda of good samaritans security session that I used up at the begginning to send the people off in the direction of other sessions. Which did well above average, but I wanted it to do better.
I had also Vista session on networking which was was kinda fun but, the real I thing I wanted to do and I am so happy that it really worked was a Data Mining session.
I really shouldn't say its Data Mining because the session was really writing more intelligent applications with values of Data Mining
[Dave] Is that something new you've done recently?
Yes, and it's you know what? The thing that's really cool about it, is that it's a very new way of thinking. It's something that no one has seriously done yet, but at least not yet on a big scale. But they are trying. Basically as a Developer, and I hope those that listen to this will mostly be developers.
[Dave] Mainly developers
As a Developer I know you are sick and tired of writing error handlers and more unexpected situations and may do this and that code, and exception handlers, and you kinda feel like your slaving for days writing code that hopefully will never execute.
[Dave] That is true
I've written today one thousand lines of really beautiful code and I hope it never runs
Can you see the? Can you see how?
Ok, so I kinda felt well lets finish that. And what would you say that if I told you that in the near future, which maybe in even today for some of you, that you would never have to write a single line of code that does data validation?
[Rich] Sounds intriguing
[Dave] Ok, Sounds good to me
[Rich] Is this the top rated session that you gave?
Yes it is, the most top rated, and I think it was just top rated not just because the technologies there, it works and it really can do what I've promised. Obviously there are a few other lines of code you still need to write, and a few things you need to learn, and a few things you need to experiment with, because we're not 100% sure about the technology.
One thing that definatley out there is that I've tapped some nerve, I've tapped something that really developers responded to. You know I've never had such, sorry to bolster myself
[Rich] No thats great
I've never had such a high score, and even if I step away from it and think ok if this was anybodys session, my goodness I think this is exactly what people wanted to hear about, and I just hope that technologies delivers in the long term
[Dave] Is this a new technology from Microsoft?
[Dave] Or something, or a new way of doing things?
No, that's the really amazing thing, the technology has been with us for about 7 years since SQL 2000 we've had it. It wasn't kinda great at that time, it was really made to work properly in SQL 2005. However, nobody almost seriously has taken it to the level that developers need, the promise of Data Mining is considered something for the BI guys, the Business Intelligence people, and they use it to find out what makes you buy more milk or more coffee or more yoghurt. But nobody thought that discovering trends and patterns in data input immeditaley highlights input that is incorrect. Isn't that cool?
[Dave] Yes, that's cool, that is well cool
Anybody who listening to it and really like wants to get more on that, obviously if you can gets you hands on the presentation that was BIN301 TechEd 2007 Barcelona by Rafal, or get on line www.sqlserverdatamining.com.
Hopefully thats easy to understand, and if you couldn't care less for reading about it just get yourself a demo. It's called DataValidation.zip and instructions are inside. It'll work on pretty much any version of Visual Studo since 2003. I just hope that it makes like 3 or 4 sessions about it.
So if any of you listening to this actually use this technology and think it's great and have some good examples you'd like to talk about and would like a little bit of advertising for you and your company get in touch with me or the Podcat managers. And if it's rubbish, tell me to!
[Dave] Ok, that's good. So, that's obviously your favourite session.
[Rich] Can you give us an example of this with example of Data Mining for the people listening?
Yeah, sure. So imagine you a data entry screen on an insurance application thing and people are entering the name of the person, age, date of birth, gender, occupation and relationship with the head of the household. So first of all, an obvious one, which is a kinda of a 0 1 case, is that if you say that your are a son and the you mark the relationship with the head of the household as daughter, that probably isn't right.
So obviously anybody will code for that with some kind of routine, well actually you don't have to. Data Mining will spot that in the past accepted transactions, so this is a little problem, you have to have some past, you have to have some logs. But that's actually not as tough as it sounds, because most applications are not green field, but brown field. If you are totally green field, you need to start obviously some point, and you need to create that log.
But if you are in a brown field redevelopment you have that data, and you are mining a pattern and the pattern will tell you straight away that if you are a male and you select that you are a daughter to the head of the household, in fact the probability of this case being a valid one is? What do we think?
Well actually no, it is 0.00003 of a percent it has to be around relatively correct, and in fact that could happen, it's just extremely unlikely. Alright, but perhaps thats not a good example because you could code a rule for that.
[Dave] That's great, I can see that because it stops you from constantly having to think the exceptions to your validation as opposed to we all write validation and then somethings wrong.
But let me take further, can I take it a little further?
[Dave/Rich] Yeah, yeah
The next step is something that not obvious. For example some one enters that the person in question is 72 years old and they are aprofessional working nurse. That is of course possible, that is absolutley so, but it is probably not very likely.
So the data entry clerks simply selected the wrong box, and what the data mining thing will now do is say ok this is possibly, but case probability is only 3%. So what you do now is set yourself a very simple threshold, so that basically there is only one line of code, which is if case probabilty is greater than the your threshold.
If you set you threshold at say 10%, which is a very high threshold, at that point you will be flagged I'm not excepting that, or at least you can say there suspicious data entry in field this and that. And you can set multiple thresholds if you want, and thats get cooler, because now we're spotting fields that you're looking, at that you've never even thought had correlations right?
[Dave] Fantastic, tell you what Rafal, run the session again quickly and we're come along. I'm intrigued already, that's brilliant, I'm very interested because we are a company of BI expertise. So I know you're saying it's not a BI...
I didn't know that
[Dave] I'm not a BI expert, but we've got guy at our company who is a BI specialist and our company does it. I'll be going back and saying, right we could use this in other ways.
May I suggest that next time you have one of the Next Generation UG group meetings that perhaps you'd like someone to actually come and talk about it and do a code demo.
[Dave] Well, maybe. How about that? Is that an offer Rafal?
That is an absolute firm and standing offer whenever you want as long as I'm available.
Can I have a PO number.
[Dave] Sure, talk to John he's out there. That's fantastic, any other sessions you've done? You said you did the Vista and the networking.
Yeah, so, looking at the score board at the moment I didn't do so well, i think only the third one from the top is my session on security, and you know the things I love to talk about, the next generation of security, CNG, SweetPea, difer helman elipitical curves, and all those lovely things I want people to understand, like finite fields over belliyam groups and so on.
[Rich] So have you been to any other sessions?
This has is an amazing TechEd for me, I've managed to attend 3 sessions, all in one time slot. So actually a third of it.
[Rich] 3 in one time slot *laughs*, and you did a speaker idol didn't you? Judging on the final Speaker Idol panel, what did you think of that?
Yeah, I feel bad about it actually. I feel bad because all the guys who came through in the end were really good.
[Rich/Dave] yeah, they were
And if I can say that we really shouldn't of given the prize to just one guy, because at least the top few were excellent and I really wanted to give them to all of them, so if anyones listening and if you are thinking of participating in Speaker Idol in the future, please do it, but please also put some pressure on these guys to make sure they open up the top prize just not to one person but to
[Dave] Yeah, it's a good idea, it's so hard to choose. We were chatting about this earlier, I think it was when Bill Aires was in full flow and i Just took a look around and there was like the judges like yourselves, all there looking like that. 200, 300 people all there, the cameras in your face, and the lights were on and I thought, these guys, I couldn't do it under the pressure. I mean how much pressure was that, it was awesome.
[Dave] They did so well! It's a great great event. I enjoyed it, thanks for joining in, it helped the whole thing.
Thank you for inviting me, if I have a chance I'll definatley do it again. I would certainly like to feedback to you at some stage a few ideas that would make it easier on the guys, because I think it was really tough on the participants.
And I don't mean easy in the sense that we shouldn't have some fun with them, the banter and all of that was very good. I kinda feel I, I and some of the other judges certainly should of, given the chance, given them some more coaching perhaps.
[Rich] Yes, I think that everyone should of, perhaps if you get through to the round you get half an hour.
[Dave] The regular speakers get some time with Rich Cleese, so i think that whoever gets to the final should qualify with an hour with Richard Cleese or something like that, or half an hour. I think that would be great for them, and I think Debbie has picked up on it, so that might happen next year.
Richard Cleese. Wow, the amazing Richard Cleese, who coaches Steve Ballmer. I just had my moment of greatness with Richard Cleese, and I'm just in awe.
That guy is amazing, he has nailed the things I do well, he has nailed the one thing that, I do apparently very well in UK, very well in Europe, but which wouldn't work for me in the States. I've been battling with how to speak in the states, I couldn't figure out what is it, and he just brought it out, at one point I know what I'm going to work on. And I hope I can start conquering the other part of the world.
[Dave] No, it was good actually here because I went for a session with him, it was only a half hour session with him last night, and I took my latop in with my demos that weren't working, so I said I'd just talk for the first 10 minutes of the slides. So I stood up and did that, and he said he didn't want to tell me too much as I was doing my session tomorrow, so there is no way you can change anything, but maybe try this and that. I thought well, some of the stuff he's saying I wouldn't bother with in a User Group environment, as you can get away with murder in a User Group environment.
I think that you're very professional in the User Group meetings I've seen, the level of professionalism everybody that was showing the presentations was incredible.
[Dave] We hopefully try and do that. But he said one or two things that were more specificly about TechEd and presentating at TechEd. I'm still not quite sure I understood him, but he said that with TechEd you should exude power, and not enthusisam, I thought, I didn't quite get what he said.
Which TechEd was that?
[Dave] That was last night with him.
I'll have to think about it
[Dave] I'm still not quite sure I got what he means. But I think I tried doing it, and it just seemed to work. There were some more other bits that he pointed out, some actual more technical bits in terms of how you lay a session out and he sort of covered that very well.
I think the stress is on exude. I think that if you exude power, you don't show power, but you show Enthusiasm. I have to check that out, that's very deep.
[Dave] Yes, it was a bit deep, yeah I thought so. Anyway, I think we've taken up enough Time, Rafal, thanks very much for a great little chat there.
[Rich] What a great way to finish TechEd!
[Rich/Dave] Thank you