Id
stringlengths
1
6
PostTypeId
stringclasses
7 values
AcceptedAnswerId
stringlengths
1
6
ParentId
stringlengths
1
6
Score
stringlengths
1
4
ViewCount
stringlengths
1
7
Body
stringlengths
0
38.7k
Title
stringlengths
15
150
ContentLicense
stringclasses
3 values
FavoriteCount
stringclasses
3 values
CreationDate
stringlengths
23
23
LastActivityDate
stringlengths
23
23
LastEditDate
stringlengths
23
23
LastEditorUserId
stringlengths
1
6
OwnerUserId
stringlengths
1
6
Tags
list
755
2
null
726
56
null
I don't know about famous, but the following is one of my favourites: > Conducting data analysis is like drinking a fine wine. It is important to swirl and sniff the wine, to unpack the complex bouquet and to appreciate the experience. Gulping the wine doesn’t work. -Daniel B. Wright (2003), see [PDF of Ar...
null
CC BY-SA 3.0
null
2010-07-27T09:44:12.930
2013-02-25T00:40:34.507
2013-02-25T00:40:34.507
183
183
null
757
2
null
73
4
null
I find [lattice](http://cran.r-project.org/web/packages/lattice/index.html) along with the companion book "Lattice: Multivariate Data Visualization with R" by Deepayan Sarkar invaluable.
null
CC BY-SA 2.5
null
2010-07-27T09:51:19.900
2010-07-27T09:51:19.900
null
null
439
null
758
2
null
723
9
null
This is a method using Monte Carlo to show whether a result is correct. Our Null Hypothesis H_0 is that our dataset does not have an interesting clustering. Our alternative hypothesis H_1 is that our dataset contains an interesting clustering. Hereby we think of interesting as, more interesting than the clustering stru...
null
CC BY-SA 2.5
null
2010-07-27T10:00:23.720
2010-07-27T10:00:23.720
null
null
190
null
759
2
null
725
3
null
I am afraid there is no; during my little adventure with such data we have just converted it to a data frame form, added some extra attributes made from neighborhoods of pixels and used standard methods. Still, packages [ripa](http://cran.r-project.org/web/packages/ripa/index.html) and [hyperSpec](http://cran.r-project...
null
CC BY-SA 2.5
null
2010-07-27T10:07:42.560
2010-07-27T10:07:42.560
null
null
null
null
760
2
null
575
22
null
I am currently writing a paper in which I have the pleasure to conduct both between and within subjects comparisons. After discussion with my supervisor we decided to run t-tests and use the pretty simple `Holm-Bonferroni method` ([wikipedia](http://en.wikipedia.org/wiki/Holm%E2%80%93Bonferroni_method)) for correcting ...
null
CC BY-SA 2.5
null
2010-07-27T10:24:44.870
2010-07-27T13:08:54.103
2010-07-27T13:08:54.103
442
442
null
761
2
null
665
21
null
I like this from Steve Skienna's [Calculated Bets](http://books.google.com/books?id=UvWGgaE4ZA8C&lpg=PA86&ots=RXQyDddYcu&dq=In%20summary%2C%20probability%20theory%20enables%20us%20to%20find%20the%20consequences%20of%20a%20given%20ideal%20world%2C%20while%20statistical%20theory%20enables%20us%20to%20to%20measure%20the%2...
null
CC BY-SA 2.5
null
2010-07-27T10:26:36.600
2010-07-27T10:26:36.600
null
null
251
null
762
2
null
575
7
null
Have a look at the [multcomp](http://cran.r-project.org/web/packages/multcomp/index.html)-package and its vignette [Simultaneous Inference in General Parametric Models](http://cran.r-project.org/web/packages/multcomp/vignettes/generalsiminf.pdf). I think it should do what wan't and the vignette has very good examples a...
null
CC BY-SA 2.5
null
2010-07-27T11:04:03.853
2010-07-27T11:04:03.853
null
null
214
null
763
2
null
562
11
null
The final year of the NetFlix Prize competition (2009) seemed to me to have sharply changed the general community-wide presumption against combining multiple learning algorithms. For instance, my formal training (university courses) and later on-the-job oversight/mentoring taught us to avoid algorithm combination unle...
null
CC BY-SA 2.5
null
2010-07-27T11:04:08.207
2010-08-04T07:39:52.463
2010-08-04T07:39:52.463
438
438
null
764
1
772
null
16
6585
I have commonly heard that LME models are more sound in the analysis of accuracy data (i.e., in psychology experiments), in that they can work with binomial and other non-normal distributions that traditional approaches (e.g., ANOVA) can't. What is the mathematical basis of LME models that allow them to incorporate the...
Linear Mixed Effects Models
CC BY-SA 2.5
null
2010-07-27T11:29:20.193
2010-08-07T17:52:12.747
2010-08-07T17:52:12.747
null
445
[ "mixed-model" ]
765
2
null
283
17
null
A model is saturated if and only if it has as many parameters as it has data points (observations). Or put otherwise, in non-saturated models the degrees of freedom are bigger than zero. This basically means that this model is useless, because it does not describe the data more parsimoniously than the raw data does (an...
null
CC BY-SA 2.5
null
2010-07-27T11:56:42.250
2010-07-27T12:32:15.423
2010-07-27T12:32:15.423
442
442
null
766
2
null
270
3
null
Most statistical packages have a function to calculate the natural logarithm of the factorial directly (e.g. the lfactorial() function in R, the lnfactorial() function in Stata). This allows you to include the constant term in the log-likelihood if you want.
null
CC BY-SA 2.5
null
2010-07-27T12:10:34.937
2010-07-27T12:10:34.937
null
null
449
null
767
2
null
577
235
null
Your question implies that AIC and BIC try to answer the same question, which is not true. The AIC tries to select the model that most adequately describes an unknown, high dimensional reality. This means that reality is never in the set of candidate models that are being considered. On the contrary, BIC tries to find...
null
CC BY-SA 4.0
null
2010-07-27T12:31:57.280
2019-08-26T17:08:49.033
2019-08-26T17:08:49.033
7290
447
null
768
2
null
665
5
null
In probability theory, we are given random variables X1, X2, ... in some way, and then we study their properties, i.e. calculate probability P{ X1 \in B1 }, study the convergence of X1, X2, ... etc. In mathematical statistics, we are given n realizations of some random variable X, and set of distributions D; the proble...
null
CC BY-SA 2.5
null
2010-07-27T12:36:47.457
2010-07-27T12:36:47.457
null
null
null
null
769
1
null
null
2
585
When does data analysis cease to be statistics ? Are the following examples all applications of statistics ?: computer vision, face recognition, compressed sensing, lossy data compression, signal processing.
What types of data analysis do not count as statistics?
CC BY-SA 2.5
null
2010-07-27T12:46:48.420
2010-07-27T16:00:22.087
null
null
327
[ "mathematical-statistics" ]
770
1
null
null
6
345
During every machine learning tutorial you'll find, there is the common "You will need to know x amount of stats before starting this tutorial". As such, using your knowledge of stats, you will learn about machine learning. My question is whether this can be reversed. Can a computer science student learn statistics t...
Is it possible to use machine learning as a method for learning stats, rather than vice-versa?
CC BY-SA 2.5
null
2010-07-27T12:53:30.250
2011-02-14T23:04:02.267
2010-09-17T20:31:31.940
null
453
[ "machine-learning", "teaching" ]
771
2
null
770
-1
null
I think that learning machine learning requires only an elementary subset of statistics; too much may be dangerous, since some intuitions are in conflict. Still, the answer to the question can it be reversed is no.
null
CC BY-SA 2.5
null
2010-07-27T13:02:58.047
2010-07-27T13:02:58.047
null
null
null
null
772
2
null
764
16
null
One major benefit of mixed-effects models is that they don't assume independence amongst observations, and there can be a correlated observations within a unit or cluster. This is covered concisely in "Modern Applied Statistics with S" (MASS) in the first section of chapter 10 on "Random and Mixed Effects". V&R walk t...
null
CC BY-SA 2.5
null
2010-07-27T13:07:38.643
2010-07-27T13:07:38.643
null
null
5
null
773
2
null
283
20
null
As everybody else said before, it means that you have as much parameters have you have data points. So, no goodness of fit testing. But this does not mean that "by definition", the model can perfectly fit any data point. I can tell you by personal experience of working with some saturated models that could not predict ...
null
CC BY-SA 2.5
null
2010-07-27T13:09:54.817
2010-07-27T13:09:54.817
null
null
447
null
774
2
null
770
8
null
I really wouldn't suggest using machine learning in order to learn statistics. The mathematics employed in machine learning is often different because there's a real emphasis on the computational algorithm. Even treatment of the same concept will be different. A simple example of this would be to compare the treatm...
null
CC BY-SA 2.5
null
2010-07-27T13:14:19.160
2010-07-27T13:14:19.160
null
null
5
null
775
1
777
null
10
8324
What is the difference between operations research and statistical analysis?
Operations research versus statistical analysis?
CC BY-SA 3.0
null
2010-07-27T13:14:54.837
2014-02-04T05:06:46.563
2012-02-23T10:41:34.117
null
460
[ "operations-research" ]
776
2
null
769
3
null
This may be slightly controversial, but certainly most of the examples that you give are not statistics (some of these would more properly fall under machine learning): Statistics usually covers the scenario where you are making inferences something when you only have a subset of the data (hence the use of things like ...
null
CC BY-SA 2.5
null
2010-07-27T13:25:13.050
2010-07-27T13:25:13.050
2017-04-13T12:44:24.667
-1
5
null
777
2
null
775
13
null
Those are entire academic discplines so I do not think you can expect much more here than pointers to further, and more extensive, documentation as e.g. Wikipedia on [Operations Research](http://en.wikipedia.org/wiki/Operations_research) and [Statistics](http://en.wikipedia.org/wiki/Statistics). Let me try a personal d...
null
CC BY-SA 2.5
null
2010-07-27T13:35:21.463
2010-07-27T13:35:21.463
null
null
334
null
778
2
null
672
15
null
I'm sorry, but there seems to be some confusion here: Bayes' theorem is not up for discussion of the neverending Bayesian-[Frequentist](http://en.wikipedia.org/wiki/Probability_interpretations#Frequentism) debate. It is a theorem that is consistent with both schools of thought (given that it is consistent with Kolmogo...
null
CC BY-SA 2.5
null
2010-07-27T13:39:35.543
2011-02-02T19:09:56.450
2011-02-02T19:09:56.450
509
447
null
779
1
null
null
9
3152
I've heard that a lot of quantities that occur in nature are normally distributed. This is typically justified using the central limit theorem, which says that when you average a large number of iid random variables, you get a normal distribution. So, for instance, a trait that is determined by the additive effect of a...
Normal distribution and monotonic transformations
CC BY-SA 2.5
null
2010-07-27T13:49:41.073
2013-03-16T00:54:30.960
2013-03-16T00:54:30.960
805
463
[ "data-transformation", "normality-assumption" ]
780
2
null
775
9
null
Operations Research (OR), sometimes called "Management Science", consists of three main topics, Optimization, Stochastic Processes, Process and Production Methodologies. OR uses statistical analysis in many contexts (for example discrete event simulations) but they should not be considered the same, additionally one o...
null
CC BY-SA 2.5
null
2010-07-27T13:55:18.623
2010-07-27T19:10:25.243
2010-07-27T19:10:25.243
172
172
null
781
2
null
712
3
null
One approach would be to use Bloom filters. Check [SAFELINK](http://www.uni-due.de/soziologie/schnell_forschung_safelink_software.php) project website for [programs](http://www.uni-due.de/methods/bloom/index-en-bloom.html) in Java and Python. Paper explaining method is [here](http://dx.doi.org/10.1186/1472-6947-9-41). ...
null
CC BY-SA 3.0
null
2010-07-27T13:59:02.623
2011-04-30T18:31:33.180
2011-04-30T18:31:33.180
22
22
null
782
2
null
779
4
null
I think you missunderstood (half of) the use statistician make of the normal distribution but I really like your question. I don't think it is a good idea to assume systematically normality and I admit it is done sometime (maybe because the normal distribution is tractable, unimodal ...) without verification. Hence y...
null
CC BY-SA 2.5
null
2010-07-27T14:02:37.327
2010-07-27T15:44:37.567
2010-07-27T15:44:37.567
223
223
null
783
2
null
726
53
null
> ... surely, God loves the .06 nearly as much as the .05. Can there be any doubt that God views the strength of evidence for or against the null as a fairly continuous function of the magnitude of p? (p.1277) Rosnow, R. L., & Rosenthal, R. (1989). Statistical procedures and the justification of knowledge in psychol...
null
CC BY-SA 4.0
null
2010-07-27T14:03:52.940
2022-11-23T09:59:13.570
2022-11-23T09:59:13.570
362671
442
null
784
2
null
614
5
null
[Street-Fighting Mathematics](https://mitpress.mit.edu/books/street-fighting-mathematics). The Art of Educated Guessing and Opportunistic Problem Solving by Sanjoy Mahajan from MIT. Available under a Creative Commons Noncommercial Share Alike license. Available as a free download on the MIT Press website (but not from ...
null
CC BY-SA 3.0
null
2010-07-27T14:04:13.360
2016-10-06T19:41:41.493
2016-10-06T19:41:41.493
122650
22
null
785
2
null
726
25
null
> The statistician cannot evade the responsibility for understanding the process he applies or recommends. -– Sir Ronald A. Fisher
null
CC BY-SA 2.5
null
2010-07-27T14:10:31.507
2010-12-03T04:02:46.950
2010-12-03T04:02:46.950
795
447
null
786
2
null
726
31
null
> The death of one man is a tragedy. The death of millions is a statistic. -- Kurt Tucholsky, in: Französischer Witz, 1925
null
CC BY-SA 3.0
null
2010-07-27T14:14:10.623
2015-12-05T15:54:52.310
2015-12-05T15:54:52.310
36678
460
null
787
2
null
726
6
null
> Statistics are the triumph of the quantitative method, and the quantitative method is the victory of sterility and death. ~ Hillaire Belloc in The Silence of the Sea
null
CC BY-SA 2.5
null
2010-07-27T14:20:42.650
2010-07-27T14:20:42.650
null
null
460
null
788
2
null
726
2
null
> A witty statesman said, you might prove anything by figures. ~ Thomas Carlyle, Chartism (1839) ch. 2
null
CC BY-SA 2.5
null
2010-07-27T14:25:39.673
2010-07-27T14:25:39.673
null
null
460
null
789
2
null
614
5
null
[Statistical Analysis with the General Linear Model](http://psy.otago.ac.nz/miller/index.htm#GLMBook) It covers basic linear models (ANOVA, ANCOVA, multiple regression). I can tell by personal experience that it is really really good book to get into the general framework of linear models, which are very useful in many...
null
CC BY-SA 3.0
null
2010-07-27T14:38:34.183
2014-06-20T06:34:29.530
2014-06-20T06:34:29.530
930
447
null
790
1
null
null
4
6855
I am looking at a scientific paper in which a single measurement is calculated using a logarithmic mean > 'triplicate spots were combined to produce one signal by taking the logarithmic mean of reliable spots' Why choose the log-mean? Are the authors making an assumption about the underlying distribution? That ...
Why (or when) to use the log-mean?
CC BY-SA 2.5
null
2010-07-27T14:46:49.640
2022-02-11T14:00:06.143
2022-02-11T14:00:06.143
11887
228
[ "distributions", "mean", "logarithm", "geometric-mean" ]
791
2
null
770
3
null
I really dont think so, as there are fundamental aspects in statistics that are simply overlooked in machine learning. For instance, in statistics, when fitting a model to data, the discrpeancy function that is used (e.g., G^2, RMSEA) is essential because they have different statistical properties. In machine learning,...
null
CC BY-SA 2.5
null
2010-07-27T14:52:19.113
2010-07-27T14:52:19.113
null
null
447
null
792
2
null
779
4
null
Simply CLT (nor any other theorem) does not state that every quantity in the universe is normally distributed. Indeed, statisticians often use monotonic transformations to improve normality, so they could use their favorite tools.
null
CC BY-SA 2.5
null
2010-07-27T15:03:28.010
2010-07-27T15:03:28.010
null
null
null
null
793
2
null
790
3
null
> Are the authors making an assumption about the underlying distribution? You are making an assumption whether you choose to use it or whether you choose against using it. For Power Law distributions it usually makes sense to look at the logarithms.
null
CC BY-SA 2.5
null
2010-07-27T15:04:58.067
2010-07-27T15:04:58.067
null
null
3807
null
794
2
null
726
45
null
> The subjectivist (i.e. Bayesian) states his judgements, whereas the objectivist sweeps them under the carpet by calling assumptions knowledge, and he basks in the glorious objectivity of science. I.J. Good
null
CC BY-SA 2.5
null
2010-07-27T15:05:38.800
2010-07-27T15:05:38.800
null
7620
7620
null
795
1
null
null
-3
2133
I have data compiled by someone else where score averages have been computed over time- averages range from 0-100. The original scores have negative values in many cases and the average would have been negative also, raw average ranges from -30 to 90. How is this 'normalization' accomplished? Thanks
Normalization of series
CC BY-SA 2.5
null
2010-07-27T15:08:11.493
2010-07-27T15:36:41.480
null
null
474
[ "mean" ]
796
2
null
305
38
null
I would like to oppose the other two answers based on a paper (in German) by [Kubinger, Rasch and Moder (2009)](http://www.psycontent.com/content/t5726m72644gq457/). They argue, based on "extensive" simulations from distributions either meeting or not meeting the assumptions imposed by a t-test, (normality and homogeni...
null
CC BY-SA 2.5
null
2010-07-27T15:14:22.937
2010-10-19T15:28:42.237
2010-10-19T15:28:42.237
442
442
null
797
2
null
779
6
null
Very good question. I feel that the answer depends on the whether you can identify the underlying process that gives rise to the measurement in question. If for example, you have evidence that height is a linear combination of several factors (e.g., height of parents, height of grandparents etc) then it would be natura...
null
CC BY-SA 2.5
null
2010-07-27T15:19:43.983
2010-07-27T15:25:00.930
2017-04-13T12:44:37.420
-1
null
null
798
1
862
null
118
265258
I'm interested in finding as optimal of a method as I can for determining how many bins I should use in a histogram. My data should range from 30 to 350 objects at most, and in particular I'm trying to apply thresholding (like Otsu's method) where "good" objects, which I should have fewer of and should be more spread o...
Calculating optimal number of bins in a histogram
CC BY-SA 3.0
null
2010-07-27T15:21:48.417
2022-11-17T19:53:13.150
2016-06-03T00:57:01.100
28666
476
[ "rule-of-thumb", "histogram" ]
799
2
null
764
9
null
The main advantage of LME for analysing accuracy data is the ability to account for a series of random effects. In psychology experiments, researchers usually aggregate items and/or participants. Not only are people different from each other, but items also differ (some words might be more distinctive or memorable, for...
null
CC BY-SA 2.5
null
2010-07-27T15:26:34.537
2010-07-27T15:26:34.537
null
null
447
null
800
2
null
795
1
null
The question is a bit unclear. But, perhaps the normalization you are looking for is this: Normalized Score = 100 * (Raw Score - min(Raw_score)) / (max(Raw Score) - min(Raw Score))
null
CC BY-SA 2.5
null
2010-07-27T15:27:18.890
2010-07-27T15:27:18.890
null
null
null
null
801
2
null
798
6
null
I'm not sure this counts as strictly good practice, but I tend to produce more than one histogram with different bin widths and pick the histogram which histogram to use based on which histogram fits the interpretation I'm trying to communicate best. Whilst this introduces some subjectivity into the choice of histogram...
null
CC BY-SA 4.0
null
2010-07-27T15:30:31.073
2020-01-24T17:34:43.113
2020-01-24T17:34:43.113
-1
210
null
802
2
null
779
5
null
The rescaling of a particular variable should, when possible, relate to some comprehensible scale for the reason that it helps make the resulting model interpretable. However, the resulting transformation need not absolutely carry a physical significance. Essentially you have to engage in a trade off between the viol...
null
CC BY-SA 2.5
null
2010-07-27T15:35:14.460
2010-07-27T15:35:14.460
null
null
196
null
803
2
null
795
4
null
Or more generically, `Index = ( ((RawValue - Min(Raw))/(Max(Raw)-Min(Raw)) * (Max(Out)-Min(Out) ) + Min(Out)` Where `Raw` is the input vector, `Out` is the output vector, and `RawValue` is the value in question. Srikant's answer is the same for an output range of 0 to 100. To convert back, rearrange to get: `RawValue ...
null
CC BY-SA 2.5
null
2010-07-27T15:36:41.480
2010-07-27T15:36:41.480
null
null
257
null
804
2
null
305
13
null
Of course, one could ditch both tests, and start using a Bayesian t-test (Savage-Dickey ratio test), which can account for equal and unequal variances, and best of all, it allows for a quantification of evidence in favor of the null hypothesis (which means, no more of old "failure to reject" talk) This test is very sim...
null
CC BY-SA 4.0
null
2010-07-27T15:43:51.857
2021-11-06T11:36:47.957
2021-11-06T11:36:47.957
11887
447
null
805
1
null
null
2
1738
I am attempting to calculate the standard error (SE) for the positive predictive value (PPV), negative predictive value (NPV), and diagnostic odds ratio (DOR) that I have obtained using the rates of true positives, false positives, true negatives, and false negatives in a sample. I am able to get 95% CIs but not SE. T...
How do I calculate the SE for PPV, NPV, and DOR?
CC BY-SA 2.5
null
2010-07-27T15:43:55.057
2010-08-13T13:35:40.603
2010-08-13T13:35:40.603
8
null
[ "error", "standard-error" ]
806
1
829
null
6
10198
Lately, there have been numerous questions about [normalization](http://en.wikipedia.org/wiki/Normalization_%28statistics%29) What are some of the situations where you never ever ever should normalize your data, and what are the alternatives?
When should normalization never be used?
CC BY-SA 2.5
null
2010-07-27T15:56:38.297
2010-09-27T23:40:16.273
null
null
59
[ "normality-assumption" ]
807
2
null
769
1
null
A statistic is any sample estimation of a population characteristic is it not? Thus, I would suggest so long as inference from a characterized sample to a population is taking place what is occurring is, at least in part, statistics. Under my definition machine learning would be a discipline that makes use of statist...
null
CC BY-SA 2.5
null
2010-07-27T16:00:22.087
2010-07-27T16:00:22.087
null
null
196
null
808
2
null
726
10
null
> A statistical analysis, properly conducted, is a delicate dissection of uncertainties, a surgery of suppositions. -- M.J. Moroney
null
CC BY-SA 2.5
null
2010-07-27T16:04:42.707
2010-10-02T17:12:07.793
2010-10-02T17:12:07.793
795
482
null
809
2
null
726
3
null
> An argument over the meaning of words is a matter of law, an argument grounded in empirical data and quantitative estimates is an argument about science. ~ Razib Khan (though he is not a statistician or famous)
null
CC BY-SA 2.5
null
2010-07-27T16:14:26.303
2010-07-27T16:14:26.303
null
null
null
null
810
2
null
798
26
null
If you use too few bins, the histogram doesn't really portray the data very well. If you have too many bins, you get a broken comb look, which also doesn't give a sense of the distribution. One solution is to create a graph that shows every value. Either a dot plot, or a cumulative frequency distribution, which doesn't...
null
CC BY-SA 3.0
null
2010-07-27T16:17:43.867
2012-12-23T06:12:04.710
2012-12-23T06:12:04.710
25
25
null
811
2
null
726
15
null
> It is the mark of a truly intelligent person to be moved by statistics. George Bernard Shaw
null
CC BY-SA 2.5
null
2010-07-27T16:18:28.013
2010-07-27T16:18:28.013
null
null
null
null
812
1
904
null
11
3856
There are a lot of references in the statistic literature to "functional data" (i.e. data that are curves), and in parallel, to "high dimensional data" (i.e. when data are high dimensional vectors). My question is about the difference between the two type of data. When talking about applied statistic methodologies tha...
What is the difference between functional data analysis and high dimensional data analysis
CC BY-SA 2.5
null
2010-07-27T16:19:42.880
2019-12-03T03:18:34.650
2019-12-03T03:18:34.650
11887
223
[ "data-mining", "curve-fitting", "signal-processing", "functional-data-analysis", "wavelet" ]
813
2
null
726
97
null
> Statistical thinking will one day be as necessary a qualification for efficient citizenship as the ability to read and write. --H.G. Wells
null
CC BY-SA 2.5
null
2010-07-27T16:23:12.707
2010-07-27T16:23:12.707
null
null
null
null
814
2
null
726
9
null
> These days the statistician is often asked such questions as "Are you a Bayesian?" "Are you a frequentist?" "Are you a data analyst?" "Are you a designer of experiments?". I will argue that the appropriate answer to ALL of these questions can be (and preferably should be) "yes", and that we can see w...
null
CC BY-SA 2.5
null
2010-07-27T16:26:53.157
2010-07-27T16:26:53.157
null
null
null
null
815
2
null
779
5
null
I must admit that I do not really understand your question: - your raindrops example is not very satisfying since this is not illustrating the fact that the Gaussian behaviour comes from the "average of a large number of iid random variables". - if the quantity $X$ that you are interested in is an average $\frac{Y_1+...
null
CC BY-SA 2.5
null
2010-07-27T16:34:22.917
2010-07-27T16:34:22.917
null
null
368
null
816
2
null
806
4
null
Of course one should never try to blindly normalize data if the data does not follow a (single) normal distribution. For example one might want to rescale observables $X$ to all be normal with $(X-\mu)/\sigma$, but this can only work if the data is normal and if both $\mu$ and $\sigma$ are the same for all data points ...
null
CC BY-SA 2.5
null
2010-07-27T16:34:55.863
2010-09-27T23:40:16.273
2010-09-27T23:40:16.273
159
56
null
817
2
null
779
5
null
Vipul, you're not being totally precise in your question. > This is typically justified using the central limit theorem, which says that when you average a large number of iid random variables, you get a normal distribution. I'm not entirely sure this is what you're saying, but keep in mind that the raindro...
null
CC BY-SA 2.5
null
2010-07-27T16:37:40.463
2010-07-27T16:37:40.463
null
null
62
null
818
2
null
73
2
null
We mostly use: - ggplot - for charts - stats - e1071 - for SVMs
null
CC BY-SA 2.5
null
2010-07-27T16:37:42.413
2010-08-09T08:15:05.380
2010-08-09T08:15:05.380
509
480
null
819
2
null
805
3
null
The PPV and NPV are proportions. You know the numerator and denominator, which both are positive integers, so you can calculate the proportion. Your goal, I presume, is to quantify how well you have determined those values. If your sample size is huge, then those values are likely to be very close to their true populat...
null
CC BY-SA 2.5
null
2010-07-27T16:40:37.527
2010-07-27T16:40:37.527
2017-03-09T17:30:36.287
-1
25
null
820
1
822
null
0
140
I am building a web application for used book trading and I am adding a feature to propose other book that would be interesting when they view an offer. Currently the data that I store are the following (they are updated each time someone visit an offer) : ISBN (book of the offer) SessId (a unique id that everyone has ...
How can I link item by relevance?
CC BY-SA 2.5
null
2010-07-27T16:42:39.070
2010-08-07T17:52:01.997
2010-08-07T17:52:01.997
null
488
[ "algorithms" ]
821
2
null
820
1
null
You probably want to look at [recommender systems](http://en.wikipedia.org/wiki/Recommender_system).
null
CC BY-SA 2.5
null
2010-07-27T16:46:45.040
2010-07-27T16:46:45.040
null
null
null
null
822
2
null
820
2
null
There are many,many ways to do this....I'd suggest googling for some search terms to get started, such as "market basket analysis", or having a look at Toby Segaran's "Programming Collective Intelligence" if you know python (even if you don't - it is pretty easy to understand).
null
CC BY-SA 2.5
null
2010-07-27T16:50:45.797
2010-07-27T16:50:45.797
null
null
247
null
823
2
null
726
-3
null
> Let’s be clear: the work of science has nothing whatever to do with consensus. Consensus is the business of politics. Science, on the contrary, requires only one investigator who happens to be right, which means that he or she has results that are verifiable by reference to the real world. In science...
null
CC BY-SA 2.5
null
2010-07-27T16:50:51.430
2010-07-27T16:50:51.430
null
null
25
null
824
2
null
672
25
null
Bayes' theorem is a relatively simple, but fundamental result of probability theory that allows for the calculation of certain conditional probabilities. Conditional probabilities are just those probabilities that reflect the influence of one event on the probability of another. Simply put, in its most famous form, i...
null
CC BY-SA 2.5
null
2010-07-27T16:56:35.273
2010-07-27T17:15:57.083
2017-03-09T17:30:36.300
-1
39
null
825
1
827
null
28
9273
I have R-scripts for reading large amounts of csv data from different files and then perform machine learning tasks such as svm for classification. Are there any libraries for making use of multiple cores on the server for R. or What is most suitable way to achieve that?
Any suggestions for making R code use multiple processors?
CC BY-SA 2.5
null
2010-07-27T17:04:37.710
2010-10-14T18:01:00.103
2010-10-14T18:01:00.103
930
480
[ "r", "parallel-computing", "multicore" ]
826
2
null
643
5
null
I like to demonstrate sampling variation and essentially the Central Limit Theorem through an "in-class" exercise. Everybody in the class of say 100 students writes their age on a piece of paper. All pieces of paper are the same size and folded in the same fashion after I've calculated the average. This is the populati...
null
CC BY-SA 3.0
null
2010-07-27T17:06:17.613
2011-06-02T08:09:07.907
2011-06-02T08:09:07.907
2116
215
null
827
2
null
825
16
null
If it's on Linux, then the most straight-forward is [multicore](http://cran.r-project.org/web/packages/multicore/index.html). Beyond that, I suggest having a look at [MPI](http://www.stats.uwo.ca/faculty/yu/Rmpi/) (especially with the [snow](http://cran.r-project.org/web/packages/snow/index.html) package). More genera...
null
CC BY-SA 2.5
null
2010-07-27T17:06:30.563
2010-07-27T17:06:30.563
null
null
5
null
828
2
null
825
5
null
Shane is correct. Both [multicore](http://cran.r-project.org/package=multicore) and [Rmpi](http://cran.r-project.org/package=Rmpi) are winners. Slightly broader coverage of the topic is in the [CRAN Task View on High-Performance Computing](http://cran.r-project.org/web/views/HighPerformanceComputing.html). This also l...
null
CC BY-SA 2.5
null
2010-07-27T17:12:15.030
2010-07-27T17:12:15.030
null
null
334
null
829
2
null
806
3
null
Whether one can normalize a non-normal data set depends on the application. For example, data normalization is required for many statistical tests (i.e. calculating a z-score, t-score, etc.) Some tests are more prone to failure when normalizing non-normal data, while some are more resistant ("robust" tests). One le...
null
CC BY-SA 2.5
null
2010-07-27T17:15:37.367
2010-07-27T17:15:37.367
null
null
491
null
830
2
null
825
3
null
Both Shane and Dirk's responses are spot on. Nevertheless, you might wanna take a look at a commercial version of R, called [Revolution R](http://www.revolutionanalytics.com/) which is built to deal with big datasets and run on multiple cores. This software is free for academics (which might be your case, I dont know...
null
CC BY-SA 2.5
null
2010-07-27T17:18:36.487
2010-07-27T17:18:36.487
null
null
447
null
831
2
null
790
1
null
In the case of log-normally distributed data .... the geometric mean is a better measure of central tendancy than the arithmetic mean. I mean I would guess they look at the paper and have seen a log-normal distribution. Spots ... makes me think its referring to probes from a microarray .. in which case they do tend to...
null
CC BY-SA 2.5
null
2010-07-27T17:19:27.563
2010-07-27T17:19:27.563
null
null
494
null
832
2
null
643
2
null
If you use Stata, you can use the -clt- command that creates graphs of sampling distributions, see [http://www.ats.ucla.edu/stat/stata/ado/teach/clt.htm](http://www.ats.ucla.edu/stat/stata/ado/teach/clt.htm)
null
CC BY-SA 2.5
null
2010-07-27T17:37:35.300
2010-07-27T17:37:35.300
null
null
null
null
833
2
null
812
18
null
Yes and no. At the theoretical level, both cases can use similar techniques and frameworks (an excellent example being Gaussian process regression). The critical difference is the assumptions used to prevent overfitting (regularization): - In the functional case, there is usually some assumption of smoothness, in othe...
null
CC BY-SA 2.5
null
2010-07-27T17:44:35.993
2010-07-28T11:37:58.753
2010-07-28T11:37:58.753
495
495
null
834
1
null
null
10
13137
I've heard that AIC can be used to choose among several models (which regressor to use). But i would like to understand formally what it is in a kind of "advanced undergraduated" level, which I think would be something formal but with intuition arising from the formula. And is it possible to implement AIC in stata wit...
What is AIC? Looking for a formal but intuitive answer
CC BY-SA 4.0
null
2010-07-27T17:46:25.223
2019-09-12T13:05:52.073
2019-09-12T13:05:52.073
11887
498
[ "stata", "aic", "intuition" ]
835
2
null
726
12
null
[Data is the sword of the 21st century, those who wield it well, the Samurai.](http://googleblog.blogspot.com/2009/02/from-height-of-this-place.html)
null
CC BY-SA 2.5
null
2010-07-27T17:57:48.807
2010-07-27T17:57:48.807
null
null
62
null
836
2
null
834
4
null
It is a heuristic, and as such, has been subjected to extensive testing. So when to trust it or not is not simple clear-cut and always-true decision. At a rough approximation, it trades off goodness of fit and number of variables ("degrees of freedom"). Much more, as usual, [at the Wikipedia article about AIC](http://...
null
CC BY-SA 2.5
null
2010-07-27T18:14:44.467
2010-07-27T18:14:44.467
null
null
334
null
837
1
858
null
2
7702
I am creating multiple logistic regression models using lrm from Harrell's Design package in R. One model I would like to make is the model with no predictors. For example, I want to predict a constant c such that: ``` logit(Y) ~ c ``` I know I how to compute c (divide the number of "1"s by the total), what I would...
R lrm model with no predictors
CC BY-SA 3.0
null
2010-07-27T18:25:40.277
2013-08-09T00:48:47.660
2013-08-09T00:47:53.890
7290
501
[ "r", "logistic" ]
838
2
null
834
5
null
Basically one needs a loss function in order to optimize anything. AIC provides the loss function which when minimized gives a "optimal"* model which fits the given data. The AIC loss function (2k-2*log(L)) tries to formulate the bias variance trade off that every statistical modeler faces when fitting a model to finit...
null
CC BY-SA 2.5
null
2010-07-27T18:28:36.730
2010-07-27T18:28:36.730
null
null
288
null
839
2
null
743
5
null
Though it is not generally labeled as Bayesian search theory, such methods are pretty widely used in oil exploration. There are, however, important differences in the standard examples that drive different features of their respective modeling problems. In the case of lost vessel exploration (in Bayesian search theo...
null
CC BY-SA 2.5
null
2010-07-27T18:33:51.930
2010-07-27T18:39:12.480
2010-07-27T18:39:12.480
39
39
null
840
2
null
73
2
null
I work with both R and [MATLAB](http://en.wikipedia.org/wiki/MATLAB) and I use [R.matlab](http://cran.r-project.org/web/packages/R.matlab/index.html) a lot to transfer data between the two.
null
CC BY-SA 2.5
null
2010-07-27T18:33:56.767
2010-08-09T08:13:30.967
2010-08-09T08:13:30.967
509
288
null
841
1
844
null
17
6126
I have $N$ paired observations ($X_i$, $Y_i$) drawn from a common unknown distribution, which has finite first and second moments, and is symmetric around the mean. Let $\sigma_X$ the standard deviation of $X$ (unconditional on $Y$), and $\sigma_Y$ the same for Y. I would like to test the hypothesis $H_0$: $\sigma_X ...
Comparing the variance of paired observations
CC BY-SA 2.5
null
2010-07-27T18:38:02.297
2011-03-13T20:31:13.040
2011-02-10T07:50:44.253
223
30
[ "distributions", "hypothesis-testing", "standard-deviation", "normal-distribution" ]
842
2
null
825
5
null
I noticed that the previous answers lack some general HPC considerations. First of all, neither of those packages will enable you to run one SVM in parallel. So what you can speed up is parameter optimization or cross-validation, still you must write your own functions for that. Or of course you may run the job for dif...
null
CC BY-SA 2.5
null
2010-07-27T18:39:32.153
2010-07-27T18:39:32.153
null
null
null
null
843
2
null
798
3
null
If I need to determine the number of bins programmatically I usually start out with a histogram that has way more bins than needed. Once the histogram is filled I then combine bins until I have enough entries per bin for the method I am using, e.g. if I want to model Poisson-uncertainties in a counting experiment with ...
null
CC BY-SA 2.5
null
2010-07-27T18:47:25.880
2010-07-27T18:47:25.880
null
null
56
null
844
2
null
841
5
null
You could use the fact that the [distribution of the sample variance](http://en.wikipedia.org/wiki/Variance#Distribution_of_the_sample_variance) is a chi square distribution centered at the true variance. Under your null hypothesis, your test statistic would be the difference of two chi squared random variates centered...
null
CC BY-SA 2.5
null
2010-07-27T18:48:54.927
2010-07-27T18:48:54.927
null
null
null
null
845
2
null
841
7
null
If you want to go down the non-parametric route you could always try the squared ranks test. For the unpaired case, the assumptions for this test (taken from [here](http://www.stat.wvu.edu/~xiawang/courses/stat551/hwk/hw14.pdf)) are: - Both samples are random samples from their respective populations. - In addition ...
null
CC BY-SA 2.5
null
2010-07-27T19:02:09.860
2010-07-28T00:08:20.203
2010-07-28T00:08:20.203
8
8
null
846
1
851
null
2
553
I'm working on regression models in STATISTICA application and I need to know what is Fisher-Snedecor distribution for and how to analyze my regression model in this distribution. What the significance level means? What is v1 and v2? I need an explanation and little tutorial on real data.
Regression output and Fisher-Snedecor distribution
CC BY-SA 3.0
null
2010-07-27T19:10:53.667
2014-12-01T17:25:36.057
2014-12-01T17:25:36.057
503
503
[ "regression" ]
847
1
null
null
0
148
Google Website Optimizer (GWO) is a tool provided by Google to do [A/B](http://en.wikipedia.org/wiki/A/B_testing) and [MVT](http://en.wikipedia.org/wiki/Multivariate_testing) experiments on websites. This has been an unanswered question for a long time so I thought I'd ask it here and see if I can get any help. Here is...
What statistical model is used to calculate the test results for GWO?
CC BY-SA 3.0
null
2010-07-27T19:12:53.037
2013-09-16T22:37:44.463
2013-09-16T22:37:44.463
22468
500
[ "computational-statistics", "model" ]
848
2
null
805
0
null
As an addition to Harvey's response if you have a small sample of data then there are alternatives to the standard error approximation formula shown above. I would recommend using software, one example would be to use the `binom.test` function in R, to calculate these confidence intervals rather than doing it "by hand"...
null
CC BY-SA 2.5
null
2010-07-27T19:19:46.097
2010-07-27T19:19:46.097
null
null
null
null
849
2
null
612
29
null
Principal Components Analysis (PCA) and Common Factor Analysis (CFA) are distinct methods. Often, they produce similar results and PCA is used as the default extraction method in the SPSS Factor Analysis routines. This undoubtedly results in a lot of confusion about the distinction between the two. The bottom line is...
null
CC BY-SA 2.5
null
2010-07-27T19:22:33.407
2010-07-27T19:27:45.063
2010-07-27T19:27:45.063
485
485
null
850
2
null
837
0
null
Try creating a dummy variable with so many ones as the number of rows in mtcars and put it formula instead of one.
null
CC BY-SA 2.5
null
2010-07-27T19:28:38.133
2010-07-27T19:28:38.133
null
null
null
null
851
2
null
846
5
null
The Fisher Snedecor distribution is another name for the [F-distribution](http://en.wikipedia.org/wiki/F-distribution). The F-distribution comes from the ratio of two chi-squared variables. In regression, you use the F-distribution in the [ANOVA](http://en.wikipedia.org/wiki/Analysis_of_variance) table. This table give...
null
CC BY-SA 2.5
null
2010-07-27T19:34:06.107
2010-07-27T19:34:06.107
null
null
8
null
852
1
null
null
5
1272
The coefficient on a logged explanatory variable when the dependent variable also is in log form is an elasticity (or the percentage change in the dependent variable if the explanatory variable changes by one percent). Suppose I estimate a regression without logging the dependent variable but I use a log link in a Gen...
Elasticities Using GLM
CC BY-SA 2.5
null
2010-07-27T19:52:26.970
2010-11-02T04:32:16.943
2010-11-02T04:32:16.943
159
null
[ "econometrics", "generalized-linear-model" ]
853
2
null
847
1
null
It sounds like they simply use Choice Modeling. You would use conditional logistic regression to analyze it. The people from JMP (SAS) have a nice (free) book about Design of Experiments and Choice Modeling.
null
CC BY-SA 2.5
null
2010-07-27T19:55:30.113
2010-07-27T19:55:30.113
null
null
74
null
854
2
null
192
2
null
I would use Fisher's Exact Test, even for large N. I wouldn't know why not. Any performance argument predates today's fast computers.
null
CC BY-SA 2.5
null
2010-07-27T19:58:44.973
2010-07-27T19:58:44.973
null
null
506
null
855
2
null
806
2
null
I thought this was too obvious, until I saw this [question](https://stats.stackexchange.com/questions/795/normalization-of-series)! When you normalise data, make sure you always have access to the raw data after normalisation. Of course, you could break this rule if you have a good reason, e.g. storage.
null
CC BY-SA 2.5
null
2010-07-27T20:05:20.027
2010-07-27T20:05:20.027
2017-04-13T12:44:52.277
-1
8
null