Apr 1, 2016

R Tools for Visual Studio (RTVS) now available: good news for MS-only shops

Microsoft informs in Newsletter #2 that they are looking for people who are willing to evaluate an "early access trial" version of their Visual Studio IDE for R, called RTVS (for R Tools for Visual Studio).

Based on the video, RTVS has the same four-window design as RStudio, so there's not an immediate struggle with an unfamiliar layout. David Smith's blog lists some of RTVS's current shortcomings, such as automated package support, that may or may not be a problem for you. I looked for signs that VS might facilitate the integration of R with other languages – such as C# for a front-end and R for the back-end – but not a whiff.

The greatest advantage of RTVS I can see is for IT shops that are comfortable

Mar 24, 2016

Control totals of a data.frame

When you are conducting a business analysis project with a data extract from the company's internal system, professional risk management suggests you make sure you are not missing any records or double counting any records. But you certainly don't want to look at every record. Yikes!

Auditors solve this predicament with control totals. When the sums of key fields and the numbers of records match known values, usually from some well-established "production report," it can be assumed your data "reconciles." *

What does it mean to calculate "control totals" of a general data.frame?

Mar 17, 2016

Google's New Search Algorithm Introduces Bias

Larry Magid has a technology "article" on the local radio station. I always turn up the volume when Magid comes on. Today's spot tells how Google Search going forward may be biased for you personally based on your Google-stored relationships. This might be handy sometimes. For example, when looking for a restaurant you may want results skewed toward your friends' favorites. Google calls these "private results." For other searches, "private results" could hide or demote the actual results you'd hoped to find. On his website Magid shows how to turn off the privatizing feature after each search, as well as how to remove it for all searches via your Google settings.

Magid mentions a third option: "Incognito" mode. In Incognito mode, it's as if you're not logged in to Google, in which case your bias-influencing relationships are (presumably!) not available. You can open a new Incognito window in Chrome via Ctrl-Shift-N. Here is the link to Google's instructions on how to browse Incognito-ly on various devices.

Mar 3, 2016

A horizontal scrolling code box in blogger

To display code in a blog I like to use a "code box" because I think it presents a more "professional" look. But it's not that easy with blogger.

By "code box" I mean a "window" with a monospaced font and vertical and horizontal scrolling bars as necessary. The internet search solutions I found almost worked, but not quite with blogger because the horizontal scroll bar wouldn't show up as expected. The vertical bar was there, but not the horizontal bar. Go figure.

But if you are comfortable hitting the HTML button next to Compose, that's easily fixed.

Feb 25, 2016

The making of a shiny mauc: chapter 2

This continues last week's post The making of a shiny mauc, based on Greg McNulty's mauc blog. It utilizes the RStudio interface to R, Desktop version.

Recall, the goal is to make an online shiny app that will run Greg's code using his data, all supplied in his post. Today we will address what modifications are necessary to show his first plot (below) on a web page. In a subsequent post we will see how to display all Greg's plots. After that we will see how

Feb 17, 2016

The making of a shiny mauc

When an excess of loss (XOL) reinsurance pricing actuary has only indemnity to work with, how can s/he reflect allocated loss adjustment expense (ALAE) in final cost projections? Such is the situation addressed by Greg McNulty in his blog Modeling ALAE Using Copulas (MAUC). According to McNulty, the classical approach — loading the indemnity value of each claim with an average ALAE/indemnity ratio — rests on "two very strong implicit assumptions": 1) ALAE and indemnity are "scaled copies" of each other and 2) ALAE and indemnity are "100% correlated." When those assumptions are questionable McNulty suggests an alternative approach.

Oct 21, 2015

Water Damage at Cal!

Dag Lohmann gave an exhaustive (not "-ing"!) presentation on Katrisk modeling of water damage at  The Berkeley R Language Beginner Study Group last night in The Ark. A bit beyond "beginner", Dag! :-) Thanks for the snips of your shiny. Nice graphs.