rfergie

13 ms·

91.
by rfergie 13y ago
I agree with what you say about the estimation of costs. The benefits to site owners normally take place outside of Facebook and so should be measurable regardless of what Facebook do. In the case where the benefit is within Facebook then t
92.
by rfergie 13y ago
That is true. If you only evaluate the effectiveness of advertising based on the number of clicks you get then you will be unable to set bids to take fraud into account.
93.
by rfergie 13y ago
Fair point. As the adverts are sold in an auction, shouldn't the cost per click reflect the proportion of fraud? E.g. if value of true click is $1 and 50% of clicks are fraudulent then CPC should surely approach $0.50?
94.
by rfergie 13y ago
I'm getting fed up with these post saying "Media X" is bad (where X is usually something to do with Facebook or Google display). Two comments: 1. This media is sold in an auction. If the quality of the traffic vs what you pay
95.
by rfergie 13y ago
It is one way of signalling "we value substance over style"
96.
by rfergie 13y ago
The HLearn libraries have a really interesting approach to this. But I think the Julia community is more active in these areas
97.
by rfergie 13y ago
I'm doing some work for a small UK based charity. I have several clustering/prediction problems in my pipeline at the moment. Drop me a line (email in profile) if you are interested in having a crack at one of them. Should give yo
98.
by rfergie 13y ago
"it should be easy" is always the killer
99.
by rfergie 13y ago
I have seen what appeared to be a penalty from pausing a campaign once. More detailed write-up at http://www.eanalytica.com/blog/unforseen-risks-when-stopping... Perhaps similar to your experience?
100.
by rfergie 13y ago
My family are not particularly conservative, but they would still be concerned about a falling GPA. Perhaps you are misdiagnosing the problem?
101.
by rfergie 13y ago
The hard part is not the coding or statistics; the hard part is figuring out what to code/analyse. I would want something on identifying actionable dimensions and how to talk to people to figure out how to help them
102.
by rfergie 13y ago
SEEKING WORK - Northern England or remote I've been working in online marketing for a few years now. Over the last few months I've had some success building my own data analysis tools (dashboards, visualisations, recommendations e
103.
by rfergie 13y ago
Perhaps it would make more sense to say "you don't have to use the same CMS for everything"
104.
by rfergie 13y ago
I'm ashamed to say that I said "in some sense" to be deliberately vague about what I meant by consistency - I didn't want to overstate the strength of Cox's theorem or end up arguing about Godel.
105.
by rfergie 13y ago
Those chapters show that any consistent (in some sense) statistical reasoning system must result in the same answers as a Bayesian system. They don't show that the frequentist approach is wrong. If both methods result in the same answe
106.
by rfergie 13y ago
A 2% bounce rate probably means that two tracking requests are being made per pageview. Common mistakes that cause this: 1. Including the tracking code twice 2. Having an "event" fire immediately on page load
107.
by rfergie 13y ago
With the recent switch over to enhanced campaigns this year I'm not even sure looking at a successful past campaign would be that great.
108.
by rfergie 13y ago
AdWords is not so simple that you can setup a successful campaign just by following a guide. Your question is like asking "Is there a complete guide for how to set up a successful business?" There are too many what-ifs for any gui
109.
by rfergie 13y ago
Don't make the mistake of thinking toolbar PR has much to do with the actual PR of a page
110.
by rfergie 13y ago
Immoral seems to be a bit of a strong word here. But then I don't use git/github much. Can you explain?
111.
by rfergie 13y ago
Just saying "machine learning" is not very useful here. What machine learning techniques work well in this case and what are the major pitfalls? Then you can convince me that a "sufficiently large training set" exists an
112.
by rfergie 13y ago
I value Gwern's style because it corresponds to the process of discovery. I think his essays are a narrative of how he comes to his conclusion. So the conclusion is less important than the journey
113.
by rfergie 13y ago
I might have something for you - will check with the boss on Monday about potential privacy issues. If I don't get back to you by Tuesday then please send me a reminder email.
114.
by rfergie 13y ago
Really you are saying "spend $1000 worth of time on this". Which might be a really great deal or not depending on how OP values their time. Out of interest, how many hours would you expect to spend on this kind of project?
115.
by rfergie 13y ago
SEEKING WORK - Northern England or Remote Expertise in web analytics and search (especially PPC but also know my way around SEO). Currently working in impact measurement in the charity sector but I have time for some other projects. Very in
116.
by rfergie 13y ago
Have you blocked javascript? All that kind of stuff disappeared for me when I installed noscript
117.
by rfergie 13y ago
Yes, I think you are right. But after reviewing my comment I think it is a point that is still worth keeping in mind. The same kind of thing could happen internally within a company or even subconsciously within a developers mind.
118.
by rfergie 13y ago
I think "King Rat" might also be part of James Clavell's series. Even if it isn't, you should like it if you like the others
119.
by rfergie 13y ago
Also be aware of the selection bias when it comes to estimations. The companies A, B and C all put in estimates for a product based on similar premises. Company C gets selected because their estimate is the lowest. This could be because com
120.
by rfergie 13y ago
I disagree that only gratuitous shit needs to be advertised. I guess you think this because you believe that if a product is good then word of mouth or whatever will mean that it sells without advertising. This is true to a certain extent i

More ›