How Fast the Fastest Human Would Run 100m?

People have used extreme value theory to predict the records in various sports. Here is an articles which provides codes to visualize the same. One can update the dataset to take into account latest records. It's interesting to see how this updation affects the estimates:

http://www.r-bloggers.com/how-fast-the-fastest-human-would-run-100m/

Checking (G)LM assumptions in R

(Generalized) Linear models make some strong assumptions concerning the data structure. Here is how to verify those assumptions in R:

http://www.r-bloggers.com/checking-glm-model-assumptions-in-r/

Some R Resources for GLMs

It is relatively easy to figure how to code a GLM in R. Even a total newcomer to R is likely to figure out that the glm() function is part of the core R language within a minute or so of searching. Thereafter though, it gets more difficult to find other GLM related stuff that R has to offer. Here is a far from complete, but hopefully helpful, list of resources.

http://www.r-bloggers.com/some-r-resources-for-glms/

Probability of Extreme Events like 9/11

An interesting article on estimating probability of large terrorist events like 9/11

http://sourish-das.blogspot.in/2013/11/probability-of-extreme-events-like-911.html

Overlapping Clusters

Aren't all of us used to seeing the well-separated clusters displayed in textbooks and papers.. But that doesn't happen in reality. So, what should one do in such cases? Read about how to deal with such situation at:

http://www.r-bloggers.com/warning-clusters-may-appear-more-separated-in-textbooks-than-in-practice/

About Normality and Testing for Normality

It is often said that with small sample sizes, everything looks normal, as the normality tests are, indeed, very sensitive to what goes on in the extreme tails. In other words, if we have enough data to fail a normality test, we always will because our real-world data won’t be clean enough. If we don’t have enough data to reliably fail a normality test, then there’s no point in performing the test, and we have to rely on the fat pencil test or our own understanding of the underlying processes. Read the detailed reasoning at:

http://www.r-bloggers.com/normality-and-testing-for-normality/

80/20 Rule of Statistical Methods Development

Developing statistical methods is hard and often frustrating work. One of the under appreciated rules in statistical methods development is the 80/20 rule. The basic idea is that the first reasonable thing you can do to a set of data often is 80% of the way to the optimal solution. Everything after that is working on getting the last 20%. The hard decision is whether to create a new method is whether the 20% is worth it. This is obviously application specific. Here is an interesting discussion about 80/20 rule of statistical methods development.

http://simplystatistics.org/2014/03/20/the-8020-rule-of-statistical-methods-development/

Bayesian First Aid

Bayesian First Aid is an attempt at implementing reasonable Bayesian alternatives to the classical hypothesis tests in R. Here are a few of them:

• Binomial Test - http://www.r-bloggers.com/bayesian-first-aid-binomial-test/
• t-test - http://www.r-bloggers.com/bayesian-first-aid-one-sample-and-paired-samples-t-test/
• Pearson Correlation Test - http://www.r-bloggers.com/bayesian-first-aid-pearson-correlation-test/

Here are a few more introductory articles:

• http://www.r-bloggers.com/an-idiot-learns-bayesian-analysis-part-1/
• http://www.r-bloggers.com/an-idiot-learns-bayesian-analysis-part-2/

Forecasting weekly data

What would you do if the seasonal period is rather long and non-integer? For example, if you have a weekly data, ARIMA models do not tend to give good results. The simplest approach in such situation is a regression with ARIMA errors. Here is an example using weekly data on US finished motor gasoline products supplied (in thousands of barrels per day) from February 1991 to May 2005.

http://www.r-bloggers.com/forecasting-weekly-data/

Photoshop via Clustering

"Do not believe anything: what artists really do is to hang around all day."
-Paco de Lucia
It seems clustering is the new way to Photoshop.. one gets different variations with different no. of clusters..
PS: Don't miss the video link in the end.

http://www.r-bloggers.com/warholing-grace-with-clara/

Movies and Statistics

It’s Oscars season again, so why shouldn't statisticians enjoy this movie fever...

Here is some number crunching with IMDb data, using R..
http://www.r-bloggers.com/predicting-movie-ratings-with-imdb-data-and-r/

Some tools on predicting Academy Awards..
http://onlinelibrary.wiley.com/doi/10.1111/j.1467-985X.2007.00518.x/abstract

A Delicious Analysis !!

A topic model is a type of statistical model for discovering the abstract "topics" that occur in a collection of documents. The article below discusses, use of Topic Modelling to find relevance of various ingredients, using data on recipes..

http://www.r-bloggers.com/a-delicious-analysis-aka-topic-modelling-using-recipes/

Princeton vs Facebook

Whoa!!! it seems two biggies here had a tussle. When Princeton claimed rapid decline in Facebook, Facebook retorted by debunking Princeton. Enjoy reading and don't forget to take away the message that how cautious one should be while doing data analysis..

http://flowingdata.com/2014/01/24/facebook-debunks-princeton-study/

Here are some third party views on this debate:

http://www.r-bloggers.com/princeton-vs-facebook-modeling-contagion/
http://www.independent.co.uk/life-style/gadgets-and-tech/facebook-is-10-whats-next-for-the-social-network-9104592.html

Linear Modeling and Logistic Regression with R

If you're new to the R language but keen to get started with linear modeling or logistic regression in the language, take a look at the link below. It works through a series of examples to teach by demonstration. All of the datasets used in the guide are available online, so it's easy to follow along from home.

http://www.r-bloggers.com/princetons-guide-to-linear-modeling-and-logistic-regression-with-r/

Recurrent events analysis, not so straightforward!

Heart failure hospitalizations are associated with an increased risk of cardiovascular death, so if an individual dies during follow-up, this isn't necessarily independent of the event process of interest. Dependent censoring needs to be accounted for in any analyses that are carried out and this renders standard methods as unsuitable. Here is some discussion about the alternative approaches:

http://www.statisticsviews.com/details/feature/5786631/Recurrent-events-analysis-not-so-straightforward.html

Charts That Don’t Start at Zero

A statistician throws light on how an improper usage of statistical tools can lead to misleading conclusions:

http://www.r-bloggers.com/lies-damn-lies-data-journalism-and-charts-that-dont-start-at-0/

Does 1+2+3… really equal -1/12?

A recent Numberphile video claims that the sum of all the positive integers is -1/12. Bothered by that, Evelyn Lamb talks about what it means to assign a value to an infinite series and explains different ways of doing this.

http://blogs.scientificamerican.com/roots-of-unity/2014/01/20/is-the-sum-of-positive-integers-negative/

Solving water resource problems using Statistics

In an exclusive interview, Dr. Upmanu Lall, Director of Columbia Water Center discusses how he uses Statistics and an understanding of climate, agriculture, commerce, engineering, technology, and politics to solve some of the world’s most pressing water problems:

http://www.statisticsviews.com/details/feature/5758681/Columbia-Professor-Uses-Statistics-and-Data-Science-to-Solve-Global-Water-Resour.html

Not Missing at Random

Not Missing at Random (NMAR) is data that is missing for a specific reason..

Here is an interesting example of NMAR data.. with the message that one shouldn't be sad and low, after reading on Facebook, about abnormally flattering lives of their friends' ..

http://simplystatistics.org/2014/01/17/missing-not-at-random-data-makes-some-facebook-users-feel-sad/

Regression with Gradient Descent

Here is an overview of the gradient descent algorithm, which offers some intuition on why the algorithm works and where it comes from, and provides examples of implementing it for ordinary least squares and logistic regression in R:

http://www.statisticsviews.com/details/feature/5722691/Getting-to-the-Bottom-of-Regression-with-Gradient-Descent.html