Teradata Aster

[learnaster]

"Data Scientists - Gifted or Made ?" https://www.linkedin.com/pulse/data-scientists-gifted-made-karthik-guruswamy … by @kguruswamy on @LinkedIn

[learnaster]

Data Scientists - Gifted or Made? What's now accepted definition of data scientist? ➥ http://bit.ly/gifted-or-made #BigData #DataScientists #DataAnalytics

[learnaster]

How Teradata Aster R Scales Data Science http://bit.ly/scales-data-science via @YouTube  

[learnaster]

Teradata Aster Analytics Going Places: On Hadoop and AWS   http://bit.ly/2bPdniU

[learnaster]

ASTER 7 - ASTER ON HADOOP IS RELEASED! http://bit.ly/2bGtXTT

[learnaster]

How do I get started with Machine Learning? Where should I begin, if I do not have a background in math? Photo Credit The Long road

[learnaster]

PART 2: Assembling the Ultimate Data Team...

by John Thuma

In this blog we will discuss these traits and what they mean and why you should attempt to avoid them.  I know it takes a village and yes this is an ideal world I am crafting but at least we have something to work toward and maybe get a laugh.  Don't get me wrong, we all have a little bit of these traits in us.  Hard to believe but I am no perfect vessel.   Lets get going!

Read more: http://bit.ly/1hZVRJO

[learnaster]

Assembling the Ultimate Data Team...

by John Thuma

I am asked by many customers about what skills should I look for when assembling the data analytic team.  The usual subjects come up: data analyst, DBA, and Data Scientist.  Those are roles and titles but something was missing from my thought process.  

So I looked back in my career and started to think about what makes a successful team.  What I came up with was a set of qualities that made the projects I was on successful.  

These qualities are "The Pioneer," "The Cattle Herder," "The Muscle," and "The Story Teller."  Crazy, I know, but give me a chance!  Many questions come up: "I can't go to HR and ask for these roles?"  "How do I use this?"  You may have people that exhibit these qualities, one person might have several of them.  

Lets dig a little deeper into each one of these types. 

Read More: http://bit.ly/1OmG41H

[learnaster]

A New Way to Understand Customer Satisfaction

by  Ryan Garrett

Our customers have found that point solutions – while perhaps satisfactory for the specific problem they are solving – provide limited development capabilities and make it difficult to factor in data sources beyond those for which they were originally designed. 

Leveraging analytic packages like R and SAS, data science teams have built incredibly insightful customer satisfaction models. But data science skills are in short supply, and executives and analysts are often left wondering about the impact of slight tweaks to the models and whether new data sources would add value – not to mention whether sampling techniques miss key activities in the data. 

Read more:http://bit.ly/1I2YsOHless

[learnaster]

Learn Machine Learning in 10 Minutes:  Naive Bayes

by John Thuma

The Naive Bayes algorithm is very simple, yet surprisingly effective. A training data set (for which we know discrete outcomes and either discrete or continuous input variables) is used to generate the model. The model is then used to predict the outcome of future observations, based on their input variables.

There are two main components of the Naive Bayes Model:

Bayes' Theroem:

Bayes’ theorem is a classical law, stating that the probability of observing an outcome given the data is proportional to the probability of observing the data given the outcome, times the prior probability of the outcome.

Naive probability Model:

The naive probability model is the assumption that the input data are independent of one another, and conditional on the outcome. This is a very strong assumption, and never true in real life, but it makes computation of all model parameters extremely simple, and violating the assumption does not hurt the model much.

Read more: http://bit.ly/1KOQHfC

[learnaster]

Aster is Machine Learning Series: History of Machine Learning

by John Thuma

Machine Learning is not a new science! Here are some pioneers and events that have helped shape our world today: The only way to determine if a machine could learn was if a human could communicate with a computer and an outside observer could not distinguish the difference between human and machine. Of course there would be the visible differences between the human and the computer would have to be overlooked. This is known as the Turing test and was proposed by Alan Turing in the 1950s. 

Read more: http://bit.ly/1GOh8M5

[learnaster]

Learn Machine Learning in 10 Minutes:  Support Vector Machines

Wonder how machines can recognize images and knows the difference between apples and oranges?

Learn Machine Learning in 10 Minutes: Support Vector Machines by John Thuma

Read more: http://bit.ly/1JXtreS

[learnaster]

Women have hearts that are physically stronger than men

by John Thuma

PROBLEM:  Find all of the instances of Ejection Fraction test results hidden in tens of millions of HL7 medical records.

WHAT IS EJECTION FRACTION:

The ejection fraction (EF) is an important measurement in determining how well your heart is pumping out blood and in diagnosing and tracking heart

failure.

A normal heart's ejection fraction may be between 55 and 70.

An EF between 40 and 55 indicates damage, perhaps from a previous heart attack, but it may not indicate heart failure.

FINDINGS:  What the team was able to determine, or correlate, was that women have hearts that are physically stronger than men.  Women also had less variance in their ejection fraction results throughout their lives.  This is what the data demonstrated and this team is not sure if there is enough evidence to prove this point.  However, it was interesting what was able to determined just be looking at HL7 text data in a statistical means.

Read more: http://bit.ly/1QktGoa

[learnaster]

Data Science - Kernel Tricks & Hyperplanes ...

by Karthik Guruswamy

To do classification - we are going to generate additional dimensions to flat data, while smartly avoiding the resulting computational complexity. It's done through an algorithm called Support Vector Machine or SVM that incorporates a cool idea called the Kernel trick. IMHO - SVM is probably one of the brilliant ideas discovered in the area of Machine Learning Supervised Classification - the reason why it's wildly successful.

The SVM model can be trained with this known data using the labels. Now, if you are given a new chapter of text from an article or book, SVM is known to predict the correct author with 95% + accuracy - Wow!

Read more: http://bit.ly/1FpW1im

[learnaster]

Data Science - Art of Dimensionality Reduction

by Karthik  Guruswamy

Just look at most of the blogs that talks about 'X ways to do Y’. The bullets together doesn't need to convey the entire information, but even if it covers 90% of insight, we've successfully reduced the dimensions of the problem.Read more: http://bit.ly/1AbBnQJ

[learnaster]

5 Qualities to Look for When Hiring a Data Scientist, 5 Qualities to Avoid

by John Thuma

The most important decision your organization will make is who it decides to hire.

In this blog we will explore the five qualities that make up a great data scientist and the five that do not.  You may be surprised by some of the attributes as they aren’t going to be what you expected.  Having a 25 year career in solution development, I have had the opportunity to work with, train, and hire many great people.  I am sure you will have a different opinion and I welcome that, but let’s get started by exploring the 5 qualities that I believe make a great data scientist.

Read more: http://bit.ly/1J6vhGA

[learnaster]