Fondue Party!

[enismirdal]

One of my flatmates is having a fondue party.

Just for general information, do NOT have an indoor fondue party unless you know what you are doing.

The aroma is...quite piquant. Other flatmate has got to work with air freshner, which amazingly enough seems to be neutralising rather than merely masking the smell; impressive!

Hey, does anyone remember if, when I have a list of experimental readings, I should give the standard deviation as sigma_n or sigma_n-1?

Votes for me to move my butt and get AIHL 3 posted before I go to bed?

Oh, and I need a coffee trip sometime in the next few days, probably to either Border's or Indigo's. Anyone wanna come?

Oh, and Voronwë the giant teddy is having a goth phase at present and currently sports a spiky belt and studded bracelets.

Comments

[crazyca.livejournal.com]

Sigma for a sum?
As far as I know, it goes by n, otherwise what's the use of the last reading if you don't include it?!

[enismirdal.livejournal.com]

Just had this feeling that with data like plant leaves, where you're only measuring a small sample of all the leaves in the world, you're meant to correct for it being justa sample by using some complicated formula that involves sigma_n-1. I don't think it actually ignores the last reading, just takes the number of groups of date to be one smaller than it is, to avoid overmeasuring... *doesn't have a clue about Stats - just dim and distant memories!*

OK, I really have no idea what I am talking about, but could use a pass-mark on this write-up! Thanks for the advice! :)

[crazyca.livejournal.com]

ROFL!
Wait! I haven't been doing the practicals for quite some time and am starting again next week!
I'm a "bit" rusty but I might actually REMEMBER what you're supposed to talk about in a week or so! :P

[fate-ofthe-free.livejournal.com]

If you still need it, it's n-1 (unless you have over 50 data points... then the two are too similar to bother with the minus 1). Since you're using the mean in calculating your standard deviation, it takes away a degree of freedom. (But if you want the standard error, divide your st deviation by root n).
Hope that helps.

[enismirdal.livejournal.com]

Wonderful! Thank you! :D

And it's not due in till Monday, so timing is fine, thanks.

[hairyears.livejournal.com]

The n-1 is there because it's a population standard deviation, not a statistic obtained by sampling the universe of all possible results.

[hairyears.livejournal.com]

...or so I'm told. On a good day, my mathematical ability extends to counting 'twenty-one' without unzipping my flies.

I have no plans for the weekend, caffeinated or otherwise. Yet.

[fate-ofthe-free.livejournal.com]

It is kind of... but that's mainly because minus one becomes negligible in a universe of all data points (since your n will be near infinite).
But in a sample, you have used all the data points to calculated the mean, so if you have this calculated mean and lose a data point, you'd be able to re-calculate this lost point from all the other points and the mean. You have lost a degree of freedom. And you're using the mean in your calculation of the standard deviation, so you must take into account this lost degree of freedom by using n-1 instead of n.

Hope that makes sense in some way or rather... I hate stats... X-(

[hairyears.livejournal.com]

That's a more accessible explanation than most of the ones I've read, although concept of a 'degree of freedom' is one that I've always found difficult.

I suspect that these things are so badly taught because of a lack of clarity of understanding (ahem) by most teachers.

Said Nile, who was on the trading floor a while ago when the new Graduate Intake were given the guided tour. They asked what I was doing, and got a brief explanation of Index Arbitrage (last year's clever trading strategy for Equity Derivatives) and one of them asked, brightly: "What's correlation?"

[psieye.livejournal.com]

Well, seems like someone's already answered the stats question, so I'll respond to this:

>> Oh, and I need a coffee trip sometime in the next few days, probably to either Border's or Indigo's. Anyone wanna come?

I guess Excel Saga or Fruits Basket won't do as an alternative to caffeine for you? ^^

[enismirdal.livejournal.com]

Would cartainly also like to catch some more Excel Saga at some point - it is highly amusing. :)

Y'see, my coffee plan at the moment is mainly something to get me into central Cambridge for some reason other than lectures, practical, supervisions or Sainsbury's - just quite fancy a chill-out with good company.

[psieye.livejournal.com]

Well, Sunday (tomorrow) seems to be semi-free to me between Chapel in the morning (10-11) and Anime Club in the afternoon (4-7). I'll call you tomorrow if I don't hear anything from you online tonight.

[enismirdal.livejournal.com]

Eeep! Tomorrow is an absolute no-no, I'm afraid; my mother is coming over! (And brings the new laptop and many other goodies; hooray!)

I'll see what my workload is like next weekend and get back to you; I'm feeling kind of drowned by the essays and write-ups at the moment!

[enismirdal.livejournal.com]

Which unfortunately means coffee plans are looking distant too. Meep!

[psieye.livejournal.com]

Mmm... next weekend will be Big Doom time again as I've got fortnightly supervisions and that weekend will be just before another triplet plus TP 1 examples class. Make plans for 2 weekends ahead then?

[psieye.livejournal.com]

Unless we're talking of after your mother leaves, i.e. evening time?

[cartesiandaemon.livejournal.com]

I can't remember the answer, but I do remember that it's obvious if you consider taking two data points... (Now I'm NotAMathmo^TM I have to remember that from school.)