Reply to: Anyone working in "Machine Learning"?

[doodab]

Just out of interest why don't you like Python?

The Zen of Python thing grates, even more so when you discover that whitespace is a fundamental part of the syntax, which appears to violate most of the aforementioned rules.

[sasguru]

The only machine learning you lot are doing is learning how to work the fryer at McDonalds.

Break the habit of a lifetime and try someything contructive for a change, why not?

[proggy]

The only machine learning you lot are doing is learning how to work the fryer at McDonalds.

[sasguru]

I'd rather be eaten alive by a python than learn it TBH

Just out of interest why don't you like Python?

[sasguru]

Edit: online version of the Barber book is here, http://web4.cs.ucl.ac.uk/staff/D.Bar...ook/090310.pdf

ta.

[doodab]

meanwhile, if you are programming oriented, you could read this book, the title seems to be a bit of a misnomer, it's about ML algorithms implemented in Python.:
Programming Collective Intelligence: Building Smart Web 2.0 Applications: Amazon.co.uk: Toby Segaran: Books
I'm not that interested in the pure programming angle, so its not for me.
There's an equivalent book for R which sadly doesn't seem as good.

This one looks good thinks its also online
http://www.amazon.co.uk/Bayesian-Rea...chine+learning

I'd rather be eaten alive by a python than learn it TBH. I have been meaning to get into octave or scilab for a while though, so I might give that a go.

Edit: online version of the Barber book is here, http://web4.cs.ucl.ac.uk/staff/D.Bar...ook/090310.pdf

[sasguru]

That's the one. It was courserras first course. It's instructor is the guy that set up courserra. Not sure when the next one runs, it seems like there is a lull in courserra courses provided over the summer, perhaps as the university professors are on summer research or what have you.

You get to learn octave. Haven't seen many jobs requiring octave. But I think the course will teach the principles which you could then apply in another language such as python

meanwhile, if you are programming oriented, you could read this book, the title seems to be a bit of a misnomer, it's about ML algorithms implemented in Python.:
Programming Collective Intelligence: Building Smart Web 2.0 Applications: Amazon.co.uk: Toby Segaran: Books
I'm not that interested in the pure programming angle, so its not for me.
There's an equivalent book for R which sadly doesn't seem as good.

This one looks good thinks its also online
http://www.amazon.co.uk/Bayesian-Rea...chine+learning

[sasguru]

I would have thought many bank jobs come with lots more stress, no?

What would be an acceptable price tag for you to do that HF job?

Bank jobs do come with stress and not very much fulfillment.
No amount of money would induce me to take that, especially when I can earn more working from home in a specialised, relatively stress free niche.

[mickey]

That sounds high pressure for not a lot of cash.

I would have thought many bank jobs come with lots more stress, no?

What would be an acceptable price tag for you to do that HF job?

[No2politics]

This one?

https://www.coursera.org/course/ml

That's the one. It was courserras first course. It's instructor is the guy that set up courserra. Not sure when the next one runs, it seems like there is a lull in courserra courses provided over the summer, perhaps as the university professors are on summer research or what have you.

You get to learn octave. Haven't seen many jobs requiring octave. But I think the course will teach the principles which you could then apply in another language such as python

[doodab]

I've heard great things about the Andrew Ng course - may look into that

This one?

https://www.coursera.org/course/ml

[DimPrawn]

Isn't this more probability than stats?

On average I'd say probably.

[doodab]

Suppose you have a normal pack of playing cards.
As usual the pack is shuffled before the games starts.
Split the pack in half.

Identify the first card in the first half of the pack, leaving 25 cards covered.
Split the other half of the pack, with 26 cards in it.

At the second card in the second half of the 26, turn it face up.

Count how many cards from this card to the first picture card.
Unless you play for a long long time you will always have an answer less than 5.
Now tell me why that is the case?
That is a nice statistical brain teaser that most people cannot fathom.

Isn't this more probability than stats?

[sasguru]

You dont have to look very far for the answer dumbo

Whatever, loser. Finished your shift at Sydney Sports Club yet?

[aussielong]

You're right - too hard for me.
Plus you're not paying me for my time, so fook orf.

HTH

You dont have to look very far for the answer dumbo