Wednesday, July 9, 2014

Review: Just Enough Math by Paco Nathan; O'Reilly Media

Not Enough Math, Simple Examples.

In this video series, Paco Nathan attempts to relate how data can be structured to take advantage of advanced math and open source tools.

Although the videos touch on some useful areas of interest, the content didn't meet my expectations with respect to explanations of the math involved and the examples are overly simplistic at less than 50 lines of python code each.

After mastering the examples, it is hard to see how a business person would make the leap to implementing data workflow based on various open source tools without knowing anything about how the underlying math works.  

On the plus side, the videos list links to additional resources (books and papers) for more details. Also, the historical context to the math concepts is given in many cases.

The videos are divided into 4 sections: Abstract Algebra, Linear Algebra, Bayesian Statistics, and Optimization. A book by the same title is due out in the summer of 2014.

In summary, my preference would be for more information regarding the math and more realistic examples.

Monday, June 16, 2014

Data Analysis Resources

A variety of resources for data analysis and machine learning. Additional suggestions are welcome.

R Package related blog posts and websites:
Books about the ggplot2 package:
Books covering the fundamentals:
Videos, Tutorial Sites, and Online Courses:
Tools:
Datasets:
  • Quandl - searchable date based numeric datasets via website or Quandl r package
Papers:

Sunday, June 1, 2014

Getting Started with Bluetooth Low Energy, by Kevin Townsend, Carles CufĂ­, Akiba, Robert Davidson, O'Reilly Media

Great for getting started.

The chapters are divided into 3 sections: Overview of BLE, Tools for Development and Testing, and Development Platforms. Depending on what you are trying to accomplish, you can jump around, but reading the book straight through is a good way to get an overall understanding before jumping into app or device development.

For app developers, there are separate chapters for Android and iOS that provide details of how to interact with BLE devices. In addition, the chapter on Application Design Tools lists several useful items that will make development much easier including: Bluetooth Application Accelerator, SensorTag, LightBlue for iOS, and nRF Master Control Panel for Android.

The last chapter, Embedded Application Development provides a starting point of what is involved with programming firmware for embedded devices.

Overall, a very good overview. I would recommend this book to anyone who is curious about the possibilities of Bluetooth LE. It seems like the combination of low cost devices and quality development tools will make for some interesting new uses.

View Getting Started with Bluetooth Low Energy in the O'Reilly Catalog.

Monday, April 21, 2014

New Game: Snow Blower App for iOS devices

Regardless of the time of year, enjoy the thrill (and challenge) of blowing snow. Clear as much snow as you can while avoiding falling icicles and a watermelon.

Available for iPad and iPhone devices from the App Store.

Snow Blower for iPhone

Snow Blower for iPad


Saturday, March 1, 2014

Mac OS X Productivity Tips for Developers, by Matthew McCullough & Tim Berglund; O'Reilly Media

This video series covers some useful tips for developers in 23 videos ranging in length from 3 to 22 minutes.

As you might expect, hotkeys, automation, and some third party apps are covered. For example, iTerm2 is presented as an improvement to the built in Terminal app. In addition, a fair bit of time is spent explaining configuration and package management tools such as HomeBrew and Boxen.  Boxen has 4 videos alone.

Personally, I liked the first videos more than the package management tool videos because there are some quick wins that can be taken advantage of right away. For instance, the different ways of using Spotlight filters could be useful for searching for things by date, etc.

The conversational delivery style is good enough. The rhetorical questions get a bit old, but I'm not sure what alternative would be better.

Also, being familiar with GitHub will make the videos easier to understand.

View Mac OS Productivity Tips for Developers in the O'Reilly Catalog.

Wednesday, February 5, 2014

Crop Insurance Calculator iPhone & iPad App

View Crop Insurance Calculator App in the iTunes App Store.

Simulate crop insurance scenarios for two common types of crop insurance: Revenue Protection and Revenue Protection with Harvest Price Exclusion.

Scenario results can be emailed.

RP = Revenue Protection Policy
RP-HPE = Revenue Protection with Harvest Price Exclusion

All fields are per acre.

APH = Actual Production History

Actual Yield = Estimated Yield (change this value to see the difference in indemnity payments)

Production Cost = Cost of production per acre. This is used to determine the Net per acre.

Premium = cost per acre of a policy.

Guarantee = Guaranteed revenue amount per acre based on the actual production history, coverage percentage, and the basis price (or harvest price).

Revenue = Actual Yield x Harvest Price. If Revenue is less than the Guarantee, then an Indemnity is paid based on the difference.

Indemnity = payment.

Net = Revenue + Indemnity - Production Cost - Premium

This application is intended as a tool for exploring various insurance options. It does not promote or advertise any insurance policies and is not affiliated with any insurance providers or the US Government.



Additional explanation of policy types from the US Risk Management Agency (RMA):

Revenue Protection policies insure producers against yield losses due to natural causes such as drought, excessive moisture, hail, wind, frost, insects, and disease, and revenue losses caused by a change in the harvest price from the projected price. The producer selects the amount of average yield he or she wishes to insure; from 50-75 percent (in some areas to 85 percent). The projected price and the harvest price are 100 percent of the amounts determined in accordance with the Commodity Exchange Price Provisions and are based on daily settlement prices for certain futures contracts. The amount of insurance protection is based on the greater of the projected price or the harvest price. If the harvested plus any appraised production multiplied by the harvest price is less than the amount of insurance protection, the producer is paid an indemnity based on the difference.

Revenue Protection With Harvest Price Exclusion policies insure producers in the same manner as Revenue Protection polices, except the amount of insurance protection is based on the projected price only (the amount of insurance protection is not increased if the harvest price is greater than the projected price). If the harvested plus any appraised production multiplied by harvest price is less than the amount of insurance protection, the producer is paid an indemnity based on the difference.

View Crop Insurance Calculator App in the iTunes App Store.

View on FarmingWithApps.com Website.

Monday, February 3, 2014

Doing Data Science: Straight Talk from the Frontline by Cathy O'Neil, Rachel Schutt; O'Reilly Media

Broadly covering the data science landscape, the authors present a wide variety of useful material originally from a graduate level Data Science course taught at Columbia University. Because the data science landscape is not well defined, the authors rely on  their personal experience (working on Google+) as well as the contributions of guest experts from various fields to fill in the texture.

Despite the complex subject matter, the writing style is easy to read and informal.

As a software engineer new to the field of data science, there were some areas where I needed to do further reading. Fortunately, the book includes lists of supplemental reading for the following subjects: math, coding, data analysis, machine learning, experimental design, and visualization.

In addition to providing details on various algorithms, the authors are careful to show some pitfalls of applying algorithms without much thought and/or using data that is not well understood or prepared.

The book includes some detailed sample code in R.

The book concludes with feedback from students, some career advice, and options for using data science skills (ex: Kaggle competitions, DataKind.org).

All in all, a very good overview.

View Doing Data Science: Straight Talk from the Frontline in the O'Reilly Catalog.