7月 112017
 

Carbon Dioxide ... CO2. Humans breathe out 2.3 pounds of it per day. It's also produced when we burn organic materials & fossil fuels (such as coal, oil, and natural gas). Plants use it for photosynthesis, which in turn produces oxygen. It is also a greenhouse gas, which many claim [...]

The post U.S. CO2 emissions are on the decline! appeared first on SAS Learning Post.

7月 112017
 

Everyone is talking about artificial intelligence (AI). In fact, many SAS customers who've been using our analytics capabilities for years or even decades are asking: What can we do with AI? What exactly is AI from a software perspective? How can we infuse cognitive computing into our customer interactions and on the customer [...]

Diary of an AI webinar was published on SAS Voices by Suzanne Clayton

7月 102017
 

In SAS Viya 3.2, SAS Visual Data Builder provides a mechanism for performing simple, self-service data preparation tasks for SAS Visual Analytics or other applications. SAS Visual Data Builder is NOT an Extract, Transform and Load (ETL) or data quality tool. You may still need one of those tools to perform more complex data preparation.

SAS Visual Data Builder can perform the following tasks:

  • View table and column profiles – provides information on number rows and columns on the table, as well as standard and advanced metrics for the columns.
  • Perform data transformations – includes items such as joining tables, transposing columns, creating calculated columns, filtering data and splitting columns.
  • Create plans – a plan is a collection of data transformations (actions) performed on one or more tables.  Plans can be saved and executed again.

SAS Visual Data Builder

To access SAS Visual Data Builder from SAS Home, select ≡ > SAS Visual Data Builder from the menu.
Note: The user must belong to the pre-defined custom user group Data Builders to have permission to access the application.

For SAS Visual Data Builder, the user can select their preferred default start screen in their application Settings.

The options are:

  • Show welcome dialog.
  • Start with data.
  • Start with new plan.
  • Choose existing plan.

With the SAS Viya 3.2 release, SAS Visual Data Builder is now a separate application from Visual Analytics (VA). There is not a one-to-one mapping of the feature set in SAS 9.4: VA 7.3 Data Preparation to SAS Viya 3.2: SAS Visual Data Builder.

For more information on SAS Visual Data Builder refer to the SAS Viya 3.2: Visual Data Builder was published on SAS Users.

7月 102017
 

I previously wrote about how to compute a bootstrap confidence interval in Base SAS. As a reminder, the bootstrap method consists of the following steps:

  1. Compute the statistic of interest for the original data
  2. Resample B times from the data to form B bootstrap samples. B is usually a large number, such as B = 5000.
  3. Compute the statistic on each bootstrap sample. This creates the bootstrap distribution, which approximates the sampling distribution of the statistic.
  4. Use the bootstrap distribution to obtain bootstrap estimates such as standard errors and confidence intervals.

In my book Simulating Data with SAS, I describe efficient ways to bootstrap in the SAS/IML matrix language. Whereas the Base SAS implementation of the bootstrap requires calls to four or five procedure, the SAS/IML implementation requires only a few function calls. This article shows how to compute a bootstrap confidence interval from percentiles of the bootstrap distribution for univariate data. How to bootstrap multivariate data is discussed on p. 189 of Simulating Data with SAS.

Skewness of univariate data

Let's use the bootstrap to find a 95% confidence interval for the skewness statistic. The data are the petal widths of a sample of 50 randomly selected flowers of the species Iris setosa. The measurements (in mm) are contained in the data set Sashelp.Iris. So that you can easily generalize the code to other data, the following statements create a data set called SAMPLE and the rename the variable to analyze to 'X'. If you do the same with your data, you should be able to reuse the program by modifying only a few statements.

The following DATA step renames the data set and the analysis variable. A call to PROC UNIVARIATE graphs the data and provides a point estimate of the skewness:

data sample;
   set sashelp.Iris;     /* <== load your data here */
   where species = "Setosa";
   rename PetalWidth=x;  /* <== rename the analyzes variable to 'x' */
run;
 
proc univariate data=sample;
   var x;
   histogram x;
   inset N Skewness (6.3) / position=NE;
run;
Distribution of petal length for 50 random Iris setosa flowers

The petal widths have a highly skewed distribution, with a skewness estimate of 1.25.

A bootstrap analysis in SAS/IML

Running a bootstrap analysis in SAS/IML requires only a few lines to compute the confidence interval, but to help you generalize the problem to statistics other than the skewness, I wrote a function called EvalStat. The input argument is a matrix where each column is a bootstrap sample. The function returns a row vector of statistics, one for each column. (For the skewness statistic, the EvalStat function is a one-liner.) The EvalStat function is called twice: once on the original column vector of data and again on a matrix that contains bootstrap samples in each column. You can create the matrix by calling the SAMPLE function in SAS/IML, as follows:

/* Basic bootstrap percentile CI. The following program is based on 
   Chapter 15 of Wicklin (2013) Simulating Data with SAS, pp 288-289. 
*/
proc iml;
/* Function to evaluate a statistic for each column of a matrix.
   Return a row vector of statistics, one for each column. */
start EvalStat(M); 
   return skewness(M);               /* <== put your computation here */
finish;
 
alpha = 0.05;
B = 5000;                            /* B = number of bootstrap samples */
use sample; read all var "x"; close; /* read univariate data into x */
 
call randseed(1234567);
Est = EvalStat(x);                   /* 1. compute observed statistic */
s = sample(x, B // nrow(x));         /* 2. generate many bootstrap samples (N x B matrix) */
bStat = T( EvalStat(s) );            /* 3. compute the statistic for each bootstrap sample */
bootEst = mean(bStat);               /* 4. summarize bootstrap distrib such as mean, */
SE = std(bStat);                             /* standard deviation,                  */
call qntl(CI, bStat, alpha/2 || 1-alpha/2);  /* and 95% bootstrap percentile CI      */
 
R = Est || BootEst || SE || CI`;     /* combine results for printing */
print R[format=8.4 L="95% Bootstrap Pctl CI"  
        c={"Obs" "BootEst" "StdErr" "LowerCL" "UpperCL"}];

The SAS/IML program for the bootstrap is very compact. It is important to keep track of the dimensions of each variable. The EST, BOOTEST, and SE variables are scalars. The S variable is a B x N matrix, where N is the sample size. The BSTAT variable is a column vector with N elements. The CI variable is a two-element column vector.

The output summarizes the bootstrap analysis. The estimate for the skewness of the observed data is 1.25. The bootstrap distribution (the skewness of the bootstrap samples) enables you to estimate three common quantities:

  • The bootstrap estimate of the skewness is 1.18. This value is computed as the mean of the bootstrap distribution.
  • The bootstrap estimate of the standard error of the skewness is 0.38. This value is computed as the standard deviation of the bootstrap distribution.
  • The bootstrap percentile 95% confidence interval is computed as the central 95% of the bootstrap estimates, which is the interval [0.49, 1.96].

It is important to realize that these estimate will vary slightly if you use different random-number seeds or a different number of bootstrap iterations (B).

You can visualize the bootstrap distribution by drawing a histogram of the bootstrap estimates. You can overlay the original estimate (or the bootstrap estimate) and the endpoints of the confidence interval, as shown below.

In summary, you can implement the bootstrap method in the SAS/IML language very compactly. You can use the bootstrap distribution to estimate the parameter and standard error. The bootstrap percentile method, which is based on quantiles of the bootstrap distribution, is a simple way to obtain a confidence interval for a parameter. You can download the full SAS program that implements this analysis.

The post Bootstrap estimates in SAS/IML appeared first on The DO Loop.

7月 072017
 

For colleges and universities, awarding financial aid today requires sophisticated analysis. When higher education leaders ask, “How can we use financial aid to help meet our institutional goals?” they need to consider many scenarios to balance strategic enrollment goals, student need, and institutional finances in order to optimize yield and [...]

Meet student enrollment goals by optimizing your financial aid strategy was published on SAS Voices by Georgia Mariani

7月 062017
 

In an IoT world, everything is connected. But what does it mean to be connected? Does it mean being plugged in to your phone, car, home, TV, favorite apps and retailers? Does it mean knowing what’s happening all around you? And having the “things” you’re connected to acting as recommender [...]

Are you getting the most out of consumer IoT data? was published on SAS Voices by Norm Marks

7月 062017
 

If there’s one thing today’s organisations can agree on, it’s that the world has changed. In the words of Tony Mooney, former managing director of insight and decision science at Sky, speaking at the recent SAS Data & Customer Experience Forum, “we are now living in a world that is volatile, uncertain, ambiguous and complex.”

As our environment has evolved, so have our techniques for understanding, measuring and motivating people and groups.

Struggling to keep up

The modern consumer is transitioning from digital-first to digital-only and they expect every business in every industry to achieve “digital parity.” In other words, your business needs to be as easy to do business with as the best of what your customer has encountered online and in self-service solutions.

 And it’s not just in the digitally native millennial generation where this transition is occurring. These changes are being realised across the generations because customers who aren’t millennials have been influenced by a millennial outlook. As a result of these changes, our expectation of brand responsibility has evolved as has our interaction with brands in a data-driven world.

Over the last decade we’ve seen a plethora of marketing technologies thrust upon us to harness this new world view and at the same time, influence it. From SMS, web analytics, mobile apps, web personalisation, recommendation engines, conversion optimisation platforms . . .the list goes on. The upshot of the investment in these myriad technologies is that it has created many disconnected silos across the organisation, each with their own set of rules and logic, focused on an individual channel and that frequently don’t speak well together. Unfortunately for consumers, the end result is a fragmented and inconsistent experience and marketers find themselves still failing to deliver a stellar customer experience.

Customer experience has been at the core of conversations about engagement for the last few years. The goal? To interact with customers with the most relevant communications at the right time via the right channel. There needs to be understanding of a customer’s attitudes preferences, interests and needs. These must be balanced with an understanding of customer lifetime value, propensity and risk to make accurate and profitable decisions about the right content, the right offer, the right price or the right product. If this can be achieved at the moment of customer engagement, then those brands won’t be the ones that get left behind.

The real-time opportunity

When we are managing outbound communications to consumers, planning email or direct marketing campaigns, we have time to consider all of the inputs from both a customer insight perspective and an internal business perspective before we decide on the most relevant content. But when a customer proactively engages with us, over the web, via an app, with a call centre agent or in person, we have just milliseconds at worst and seconds at best to make the most accurate and profitable decision. This becomes a major challenge.

According to our recent research and speaking to key decision-makers in consumer organisations, one in five believe the ability to interact with customers and adjust those interactions in real-time (based on the most up to date insight and context), would see revenues jump by as much as 20-40 percent. The majority of decision-makers expect revenues to increase by at least 10 percent.

Daragh Kelly, Data Strategy & Innovation Director at Sky, echoed the thoughts of Mooney at the forum, saying the key to achieving this is “improving all of the small decisions that are made by organisations when they interact with customers." Small decisions are those made in response to an individual customer’s choices and a focus on small decisions offers benefits through a multitude of applications:

  • Improving the management of risk and the matching of price to risk.
  • Reducing or eliminating fraud and waste.
  • Increasing revenue by making the most of every opportunity
  • Improving the utilisation of constrained resources across the organisation, all whilst delivering a superior customer experience.

Organisations need to adapt from making decisions at the speed of the organisation to making decision at the speed of the customer.  For instance, if a customer chooses not to engage with an offer online, based on all the information known about that individual including their lifetime value, propensity and attitudes, as well as new contextual information (e.g., their location, the device they’re using, etc.), they can be served a more suitable alternative within seconds.

Planning and process

To remain relevant in increasingly competitive and disruptive markets and to meet the expectations of the modern consumer, organisations need to put a framework in place to enable them to make better "small decisions" at those moment of customer interaction, which are the true "crunch" moments for individual customers.

Permanent TSB is one organisation that has started to implement such a framework. Underpinned by in depth and advanced customer analytics, the organisation has moved from engaging with customers via outbound only calling campaign structure to developing an omni-channel engagement framework. Through customer analytics, the organisation has been able to prioritise activities and deliver services, offers and updates highly tailored for individual customers.

Businesses like Allied Irish Bank have made analytics a key strategic pillar, with buy-in from the C-level down through the organisation. Customer analytics is being used to drive informed and accurate decision-making right across the business.

Consumers don’t think in channels. They just want to do business with you in a way that is easy, consistent and relevant, regardless of how and when they interact with you. This means they expect you to know them as a customer, to understand their previous engagements and transactions with you and to use the data they make available to you. Only then can you deliver a personalised and relevant experience, every time. Broad brush segmentation approaches to customer interactions based on rules and demographics will no longer cut it. Instead we need to get to a segment of one. Making analytical decisions, based on an in-depth understanding of each individual customer and making those decisions at the speed of the customer rather than at the speed of the organisation, is the key to delivering the superior customer experiences now being demanded by all consumers.

Find out more about how data analytics can deliver personalised decisions to customers in real-time.

The age of now: focusing on the segment of one was published on Customer Intelligence Blog.

7月 062017
 

If there’s one thing today’s organisations can agree on, it’s that the world has changed. In the words of Tony Mooney, former managing director of insight and decision science at Sky, speaking at the recent SAS Data & Customer Experience Forum, “we are now living in a world that is volatile, uncertain, ambiguous and complex.”

As our environment has evolved, so have our techniques for understanding, measuring and motivating people and groups.

Struggling to keep up

The modern consumer is transitioning from digital-first to digital-only and they expect every business in every industry to achieve “digital parity.” In other words, your business needs to be as easy to do business with as the best of what your customer has encountered online and in self-service solutions.

 And it’s not just in the digitally native millennial generation where this transition is occurring. These changes are being realised across the generations because customers who aren’t millennials have been influenced by a millennial outlook. As a result of these changes, our expectation of brand responsibility has evolved as has our interaction with brands in a data-driven world.

Over the last decade we’ve seen a plethora of marketing technologies thrust upon us to harness this new world view and at the same time, influence it. From SMS, web analytics, mobile apps, web personalisation, recommendation engines, conversion optimisation platforms . . .the list goes on. The upshot of the investment in these myriad technologies is that it has created many disconnected silos across the organisation, each with their own set of rules and logic, focused on an individual channel and that frequently don’t speak well together. Unfortunately for consumers, the end result is a fragmented and inconsistent experience and marketers find themselves still failing to deliver a stellar customer experience.

Customer experience has been at the core of conversations about engagement for the last few years. The goal? To interact with customers with the most relevant communications at the right time via the right channel. There needs to be understanding of a customer’s attitudes preferences, interests and needs. These must be balanced with an understanding of customer lifetime value, propensity and risk to make accurate and profitable decisions about the right content, the right offer, the right price or the right product. If this can be achieved at the moment of customer engagement, then those brands won’t be the ones that get left behind.

The real-time opportunity

When we are managing outbound communications to consumers, planning email or direct marketing campaigns, we have time to consider all of the inputs from both a customer insight perspective and an internal business perspective before we decide on the most relevant content. But when a customer proactively engages with us, over the web, via an app, with a call centre agent or in person, we have just milliseconds at worst and seconds at best to make the most accurate and profitable decision. This becomes a major challenge.

According to our recent research and speaking to key decision-makers in consumer organisations, one in five believe the ability to interact with customers and adjust those interactions in real-time (based on the most up to date insight and context), would see revenues jump by as much as 20-40 percent. The majority of decision-makers expect revenues to increase by at least 10 percent.

Daragh Kelly, Data Strategy & Innovation Director at Sky, echoed the thoughts of Mooney at the forum, saying the key to achieving this is “improving all of the small decisions that are made by organisations when they interact with customers." Small decisions are those made in response to an individual customer’s choices and a focus on small decisions offers benefits through a multitude of applications:

  • Improving the management of risk and the matching of price to risk.
  • Reducing or eliminating fraud and waste.
  • Increasing revenue by making the most of every opportunity
  • Improving the utilisation of constrained resources across the organisation, all whilst delivering a superior customer experience.

Organisations need to adapt from making decisions at the speed of the organisation to making decision at the speed of the customer.  For instance, if a customer chooses not to engage with an offer online, based on all the information known about that individual including their lifetime value, propensity and attitudes, as well as new contextual information (e.g., their location, the device they’re using, etc.), they can be served a more suitable alternative within seconds.

Planning and process

To remain relevant in increasingly competitive and disruptive markets and to meet the expectations of the modern consumer, organisations need to put a framework in place to enable them to make better "small decisions" at those moment of customer interaction, which are the true "crunch" moments for individual customers.

Permanent TSB is one organisation that has started to implement such a framework. Underpinned by in depth and advanced customer analytics, the organisation has moved from engaging with customers via outbound only calling campaign structure to developing an omni-channel engagement framework. Through customer analytics, the organisation has been able to prioritise activities and deliver services, offers and updates highly tailored for individual customers.

Businesses like Allied Irish Bank have made analytics a key strategic pillar, with buy-in from the C-level down through the organisation. Customer analytics is being used to drive informed and accurate decision-making right across the business.

Consumers don’t think in channels. They just want to do business with you in a way that is easy, consistent and relevant, regardless of how and when they interact with you. This means they expect you to know them as a customer, to understand their previous engagements and transactions with you and to use the data they make available to you. Only then can you deliver a personalised and relevant experience, every time. Broad brush segmentation approaches to customer interactions based on rules and demographics will no longer cut it. Instead we need to get to a segment of one. Making analytical decisions, based on an in-depth understanding of each individual customer and making those decisions at the speed of the customer rather than at the speed of the organisation, is the key to delivering the superior customer experiences now being demanded by all consumers.

Find out more about how data analytics can deliver personalised decisions to customers in real-time.

The age of now: focusing on the segment of one was published on Customer Intelligence Blog.