Tag Archives: Machine Learning

Sensor Analytics on Big Data at Micro Scale

We often think of analytics on large scales, particularly in the context of large data sets (“Big Data”). However, there is a growing analytics sector that is focused on the smallest scale. That is the scale of digital sensors — driving us into the new era of sensor analytics.

Small scale (i.e., micro scale) is nothing new in the digital realm. After all, the digital world came into existence as a direct consequence of microelectronics and microcircuits. We used to say in the early years of astronomy big data (which is my background) that the same transistor-based logic microcircuitry that comprises our data storage devices (which are storing massive streams of data) is essentially the same transistor-based logic microcircuitry inside our sensors (which are collecting that data). The latter includes, particularly, the sensors inside digital cameras, consisting of megapixels and even gigapixels. Consequently, there should be no surprise that the two digital data functions (sensing and storing) are intimately connected and that we are therefore drowning in oceans of data.

But, in our rush to crown data “big”, we sometimes may have forgotten that micro-scale component to the story. But not any longer. There is growing movement in the microchip world in new and interesting directions.

I am not only talking about evolutions of the CPU (central processing unit) that we have seen for years: the GPU (graphics processing unit) and the FPGA (field programmable gate array).  We are now witnessing the design, development, and deployment of more interesting application-specific integrated circuits (ASICs), one of which is the TPU (tensor processing unit) which is specifically designed for AI (artificial intelligence) applications. The TPU (as its name suggests) can perform tensor calculations on the chip, in the same way that earlier generation integrated circuits were designed to perform scalar operations (in the CPU) and to perform vector and/or parallel streaming operations (in the GPU).

Speeding the calculations is precisely the goal of these new chips. One that I heard discussed in the context of cybersecurity applications is the BPU (Behavior Processing Unit), designed to detect BOI (behaviors of interest). Whereas the TPU might be detecting persons of interest (POI) or objects of interest in an image, the BOI is looking at patterns in the time series (sequence data) that are indicative of interesting (and/or anomalous) behavioral patterns.

The BOI detector (the BPU) would definitely represent an amplifier to cybersecurity operations, in which the massive volumes of data streaming through our networks and routers are so huge that we never actually capture (and store) all of that data. So we need to detect the anomalous pattern in real-time before a damaging cyber incident occurs!

You can continue reading the long version of this article and learn more about the growing class of new analytics ASIC processors in my article “Sensor Analytics at Micro Scale on the xPU” at the Western Digital DataMakesPossible.com blog site.

Learn more about Machine Learning for Edge Devices at Western Digital here: https://blog.westerndigital.com/machine-learning-edge-devices/

Finally, see what’s cooking in Western Digital’s new Machine Learning Accelerator here: https://blog.westerndigital.com/machine-learning-accelerator-embedded-world-2019/

Brain CPU Chip for AI Acceleration

Brief Guide to xPU for AI Accelerators

Source for graphic: https://www.sigarch.org/a-brief-guide-of-xpu-for-ai-accelerators/

Meta-Learning For Better Machine Learning

In a related post we discussed the Cold Start Problem in Data Science — how do you start to build a model when you have either no training data or no clear choice of model parameters. An example of a cold start problem is k-Means Clustering, where the number of clusters k in the data set is not known in advance, and the locations of those clusters in feature space (i.e., the cluster means) are not known either. So, you start by assuming a value for k and making random assumptions about the cluster means, and then iterate until you find the optimal set of clusters, based upon some evaluation metric. See the related post for more details about the cold start challenge. See the attached graphic below for a simple demonstration of a k-Means Clustering application.

The above example (clustering) is taken from unsupervised machine learning (where there are no labels on the training data). There are also examples of cold start in supervised machine learning (where you do have class labels on the training data).

As an example of a cold start in supervised learning, we look at neural network models, where the weights on the edges that connect the various nodes in the network layers are not known initially. So, random values (e.g., all weights = 1) are assigned to all of the edge weights (which could number in the hundreds or thousands) — that’s the cold start. Following that, the weights can “learn” to get better through a technique known as backpropagation, which is applied through sequential iterations of the neural network learning process. A validation metric estimates the error in each model iteration in the sequence (i.e., the classification error on the validation or hold-out data set), then applies the backpropagation technique to assign some portion of the error to each of the edge weights. Each edge weight is adjusted accordingly using gradient descent (or some similar error correction rate estimator) for the next model in the sequence. The next iteration of the neural network modeling process is executed, applying the same steps as above, and the process continues until the validation metric converges to the optimal final model.

What is missing in the above discussion is the deeper set of unknowns in the learning process. This is the meta-learning phase. We can elucidate this phase through our two examples above.

From the first example above, k-Means Clustering:

  • What is the value of k?
  • Which features in the data set are most effective in creating distinct clusters in the data (i.e., to create the segments that are the most compact internally, and relatively the most separated from each other)? There might be dozens or hundreds or thousands of attributes to choose from, and a vast number of combinations of those attributes in which to explore clustering in different dimensions of parameter space.
  • What distance metric should be used to estimate separation (or what similarity metric should be used to estimate similarity), since clustering is a distance-based algorithm? There are some common choices for distance and similarity metrics (e.g., cosine similarity, Euclidean distance, Manhattan distance, Mahalanobis distance, Lp-Norm, etc.), but that is just the tip of a vast iceberg — just take a look at the 750-page book Encyclopedia of Distances.
  • What evaluation metric should we use to determine if the clusters are “good enough” or optimal (i.e., the most compact set of clusters relative to the separation of the clusters)? There are several choices for such evaluation metrics: Dunn index, Davies-Bouldin index, C-index, and Silhouette analysis are just a few examples.

We need to decide on all of these parameterizations of the clustering model before the cold start interations on the cluster means can begin.

From the second example above, Neural Network modeling, there are also many different preliminary tasks and parameterizations of the network that need to be decided and acted on before the cold start iterations on the edge weights can begin:

This now gets to the heart of meta-learning. It is focused on learning the right tasks to perform and on tuning the modeling hyper-parameters. These are the different tasks and “external” parameters that differentiate various instantiations of a specific model within a broader category of models — those tasks and external parameterizations must be explored before you start building, iterating, and validating a specific model’s “internal” parameters. For example:

  • You can cluster children’s toys in a toy store by color, or by shape, or by electronic vs. non-electronic, or by age-appropriateness, or by functionality, or cluster them by some combination of those features.
  • You can cluster (segment) your customers by the types of products they buy, or by their geographic location, or by their gender, or by their age, or by the day of week that they prefer to shop, or cluster them by some combination of those many different variables.
  • You can cluster medical drug treatments by the types of symptoms that they address, or by the medical diagnoses (outcomes) that they attempt to cure, or by their dosage amounts, or cluster them by the side-effects that are caused when different combinations of the drugs are used.

Deciding on the higher-level hyper-parameterizations of your clustering approach before you build the actual models is good data science and good business, no matter whether you are sorting toys, or discovering segments in your customer database, or prescribing different medications to medical patients.

Similar decisions must be made for the neural network example mentioned earlier as well as for numerous other machine learning modeling techniques. Meta-learning is important to make sure that you are aware of and attentive to the many choices of modeling tasks and parameterizations for the models that you are about to train. Meta-learning is also critical for demonstrating (proving) that you built the best (or optimal or most accurate) model, given the higher level characteristics (e.g., parameters, architecture, or input data sources) of the modeling effort:

  • What is the business case? What outcomes will be actionable?
  • What data do we have? Which combinations of data have we not explored yet?
  • What metric will demonstrate that we have achieved the globally optimal model (or approximately the global optimum), versus some locally good model that doesn’t generalize across a larger data set?

Genetic Algorithms (GAs) are an example of meta-learning. They are not machine learning algorithms in themselves, but GAs can be applied across ensembles of machine learning models and tasks, in order to find the optimal model (perhaps globally optimal model) across a collection of locally optimal solutions.

Learn more about meta-learning from these resources:

Finally, in addition to the awesome 750-page book Encyclopedia of Distances“, please check out some of these top-selling books on Data Science, AI, and Machine Learning:

Data Scientist’s Dilemma – The Cold Start Problem

The ancient philosopher Confucius has been credited with saying “study your past to know your future.” This wisdom applies not only to life but to machine learning also. Specifically, the availability and application of labeled data (things past) for the labeling of previously unseen data (things future) is fundamental to supervised machine learning.

Without labels (diagnoses, classes, known outcomes) in past data, then how do we make progress in labeling (explaining) future data? This would be a problem.

A related problem also arises in unsupervised machine learning. In these applications, there is no requirement or presumption regarding the existence of labeled training data — we are essentially parameterizing or characterizing the patterns in the data (e.g., the trends, correlations, segments, clusters, associations).

Many unsupervised learning models can converge more readily and be more valuable if we know in advance which parameterizations are best to choose. If we cannot know that (i.e., because it truly is unsupervised learning), then we would like to know at least that our final model is optimal (in some way) in explaining the data.

In both of these applications (supervised and unsupervised machine learning), if we don’t have these initial insights and validation metrics, then how does such model-building get started and get moving towards the optimal solution?

This challenge is known as the cold-start problem! The solution to the problem is easy (sort of): We make a guess — an initial guess! Usually, that would be a totally random guess.

That sounds so… so… random! How do we know whether it’s a good initial guess? How do we progress our model (parameterizations) from that random initial choice? How do we know that our progression is moving towards more accurate models? How? How? How?

This can be a real challenge. Of course nobody said the “cold start” problem would be easy. Anyone who has ever tried to start a very cold car on a frozen morning knows the pain of a cold start challenge. Nothing can be more frustrating on such a morning. But, nothing can be more exhilarating and uplifting on such a morning than that moment when the engine starts and the car begins moving forward with increasing performance.

The experiences for data scientists who face cold-start problems in machine learning can be very similar to those, especially the excitement when our models begin moving forward with increasing performance.

We will itemize several examples at the end. But before we do that, let’s address the objective function. That is the true key that unlocks performance in a cold-start challenge.  That’s the magic ingredient in most of the examples that we will list.

The objective function (also known as cost function, or benefit function) provides an objective measure of model performance. It might be as simple as the percentage of class labels that the model got right (in a classification model), or the sum of the squares of the deviations of the points from the model curve (in a regression model), or the compactness of the clusters relative to their separation (in a clustering analysis).

The value of the objective function is not only in its final value (i.e., giving us a quantitative overall model performance rating), but its great (perhaps greatest) value is realized in guiding our progression from the initial random model (cold-start zero point) to that final successful (hopefully, optimal) model. In those intermediate steps it serves as an evaluation (or validation) metric.

By measuring the evaluation metric at step zero (cold-start), then measuring it again after making adjustments to the model parameters, we learn whether our adjustments led to a better performing model or worse performance. We then know whether to continue making model parameter adjustments in the same direction or in the opposite direction. This is called gradient descent.

Gradient descent methods basically find the slope (i.e., the gradient) of the performance error curve as we progress from one model to the next. As we learned in grade school algebra class, we need two points to find the slope of a curve. Therefore, it is only after we have run and evaluated two models that we will have two performance points — the slope of the curve at the latest point then informs our next choice of model parameter adjustments: either (a) keep adjusting in the same direction as the previous step (if the performance error decreased) to continue descending the error curve; or (b) adjust in the opposite direction (if the performance error increased) to turn around and start descending the error curve.

Note that hill-climbing is the opposite of gradient descent, but essentially the same thing. Instead of minimizing error (a cost function), hill-climbing focuses on maximizing accuracy (a benefit function). Again, we measure the slope of the performance curve from two models, then proceed in the direction of better-performing models. In both cases (hill-climbing and gradient descent), we hope to reach an optimal point (maximum accuracy or minimum error), and then declare that to be the best solution. And that is amazing and satisfying when we remember that we started (as a cold-start) with an initial random guess at the solution.

When our machine learning model has many parameters (which could be thousands for a deep neural network), the calculations are more complex (perhaps involving a multi-dimensional gradient calculation, known as a tensor). But the principle is the same: quantitatively discover at each step in the model-building progression which adjustments (size and direction) are needed in each one of the model parameters in order to progress towards the optimal value of the objective function (e.g., minimize errors, maximize accuracy, maximize goodness of fit, maximize precision, minimize false positives, etc.). In deep learning, as in typical neural network models, the method by which those adjustments to the model parameters are estimated (i.e., for each of the edge weights between the network nodes) is called backpropagation. That is still based on gradient descent.

One way to think about gradient descent, backpropagation, and perhaps all machine learning is this: “Machine Learning is the set of mathematical algorithms that learn from experience. Good judgment comes experience. And experience comes from bad judgment.” In our case, the initial guess for our random cold-start model can be considered “bad judgment”, but then experience (i.e., the feedback from validation metrics such as gradient descent) bring “good judgment” (better models) into our model-building workflow.

Here are ten examples of cold-start problems in data science where the algorithms and techniques of machine learning produce the good judgment in model progression toward the optimal solution:

  • Clustering analysis (such as K-Means Clustering), where the initial cluster means and the number of clusters are not known in advance (and thus are chosen randomly initially), but the compactness of the clusters can be used to evaluate, iterate, and improve the set of clusters in a progression to the final optimum set of clusters (i.e., the most compact and best separated clusters).
  • Neural networks, where the initial weights on the network edges are assigned randomly (a cold-start), but backpropagation is used to iterate the model to the optimal network (with highest classification performance).
  • TensorFlow deep learning, which uses the same backpropagation technique of simpler neural networks, but the calculation of the weight adjustments is made across a very high-dimensional parameter space of deep network layers and edge weights using tensors.
  • Regression, which uses the sum of the squares of the deviations of the points from the model curve in order to find the best-fit curve. In linear regression, there is a closed-form solution (derivable from the linear least-squares technique). The solution for non-linear regression is not typically a closed-form set of mathematical equations, but the minimization of the sum of the squares of deviations still applies — gradient descent can be used in an iterative workflow to find the optimal curve. Note that K-Means Clustering is actually an example of piecewise regression.
  • Nonconvex optimization, where the objective function has many hills and valleys, so that gradient descent and hill-climbing will typically converge only to a local optimum, not to the global optimum. Techniques like genetic algorithms, particle swarm optimization (when the gradient cannot be calculated), and other evolutionary computing methods are used to generate lots of random (cold-start) models and then iterate each of them until you find the global optimum (or until you run out of time and resources, and then pick the best one that you could find). [See my graphic attached below that illustrates a sample use case for genetic algorithms. See also the NOTE below the graphic about Genetic Algorithms, which also applies to other evolutionary algorithms, indicating that these are not machine learning algorithms specifically, but they are actually meta-learning algorithms]
  • kNN (k-Nearest Neighbors), which is a supervised learning technique in which the data set itself becomes the model. In other words, the assignment of a new data point to a particular group (which may or may not have a class label or a particular meaning yet) is based simply upon finding which category (group) of existing data points is in the majority when you take a vote of the nearest neighbors to the new data point. The number of nearest neighbors that are to be examined is some number k, which can be initially arbitrary (a cold-start), but then it is adjusted to improve model performance.
  • Naive Bayes classification, which applies Bayes theorem to a large data set with class labels on the data items, but for which some combinations of attributes and features are not represented in the training data (i.e., a cold-start challenge). By assuming that the different attributes are mutually independent features of the data items, then one can estimate the posterior likelihood for what the class label should be for a new data item with a feature vector (set of attributes) that is not found in the training data. This is sometimes called a Bayes Belief Network (BBN) and is another example of where the data set becomes the model, where the frequency of occurrence of the different attributes individually can inform the expected frequency of occurrence of different combinations of the attributes.
  • Markov modeling (Belief Networks for Sequences) is an extension of BBN to sequences, which can include web logs, purchase patterns, gene sequences, speech samples, videos, stock prices, or any other temporal or spatial or parametric sequence.
  • Association rule mining, which searches for co-occurring associations that occur higher than expected from a random sampling of a data set. Association rule mining is yet another example where the data set becomes the model, where no prior knowledge of the associations is known (i.e., a cold-start challenge). This technique is also called Market Basket Analysis, which has been used for simple cold-start customer purchase recommendations, but it also has been used in such exotic use cases as tropical storm (hurricane) intensification prediction.
  • Social network (link) analysis, where the patterns in the network (e.g., centrality, reach, degrees of separation, density, cliques, etc.) encode knowledge about the network (e.g., most authoritative or influential nodes in the network), through the application of algorithms like PageRank, without any prior knowledge about those patterns (i.e., a cold-start).

Finally, as a bonus, we mention a special case, Recommender Engines, where the cold-start problem is a subject of ongoing research. The research challenge is to find the optimal recommendation for a new customer or for a new product that has not been seen before. Check out these articles  related to this challenge:

  1. The Cold Start Problem for Recommender Systems
  2. Tackling the Cold Start Problem in Recommender Systems
  3. Approaching the Cold Start Problem in Recommender Systems

We started this article mentioning Confucius and his wisdom. Here is another form of wisdomhttps://rapidminer.com/wisdom/ — the RapidMiner Wisdom conference. It is a wonderful conference, with many excellent tutorials, use cases, applications, and customer testimonials. I was honored to be the keynote speaker for their 2018 conference in New Orleans, where I spoke about “Clearing the Fog around Data Science and Machine Learning: The Usual Suspects in Some Unusual Places”. You can find my slide presentation here: KirkBorne-RMWisdom2018.pdf 

NOTE: Genetic Algorithms (GAs) are an example of meta-learning. They are not machine learning algorithms in themselves, but GAs can be applied across ensembles of machine learning models and tasks, in order to find the optimal model (perhaps globally optimal model) across a collection of locally optimal solutions.

Variety is the Secret Sauce for Big Discoveries in Big Data

When I was out for a walk recently, I heard a loud low-flying aircraft passing overhead. This was not unusual since we live in the flight path of planes landing at a major international airport about 10 miles from our home. In this case, I thought to myself that the sound seemed more directly overhead and lower than normal as well as being suggestive of a larger than average jet aircraft.

I realized that in my one simple thought, I had made three different inferences from a single stream of data. The data stream was the audible sound of the aircraft. The three inferences were about the altitude (lower than normal), the size (larger than average), and the flight path (more overhead). When I looked up, my tri-inference hypothesis was confirmed. The plane was a very large, low-flying jet for a major overnight shipping company. The slightly unusual flight path may have been associated with the fact that these planes are probably instructed to land on a different runway at the airport than the usual commercial passenger airlines’ flights – consequently, the altitude and location were slightly different from the slightly smaller commercial passenger airlines that pass overhead every day.

This situation caused me to reflect on how often we can jump to conclusions, infer a hypothesis, and (maybe without as much proof as in this case) we assume that our conclusion is true.

For the modern digital organization, the proof of any inference (that drives decisions) should be in the data! Rich and diverse data collections enable more accurate and trustworthy conclusions.

I frequently refer to the era of big data as “the end of demographics”. By that, I mean that we now have many more features, attributes, data sources, and insights into each entity in our domain: people, processes, and products. These multiple data sources enable a “360 view” of the entity, thus empowering a more personalized (even hyper-personalized) understanding of and response to the needs of that unique entity. In “big data language”, we are talking about one of the 3 V’s of big data: big data Variety!

High variety is one of the foundational key features of big data — we now measure many more features, characteristics, and dimensions of insight into nearly everything due to the plethora of data sources, sensors, and signals that we measure, monitor, and mine. Consequently, we no longer need to rely on a limited number of features and attributes when making decisions, taking actions, and generating inferences. We can make better, tailored, more personalized decisions and actions. Every entity is unique! That marks the end of demographics.

Here is another example: suppose that a person goes to their doctor to report problems with painful headaches. That is a single symptom (headache pain) — a single data source, a single signal, a single sensor. However, one could imagine a large number of possible inferences from that one single signal. The headaches could be caused by insufficient sleep (sleep apnea), high blood pressure, pregnancy, or a brain tumor. Obviously, each one of these diagnoses carries a seriously different course of action and treatment.

In “data science language”, what we are describing are different segments (clusters) in the hyperspace of symptoms and causes in which the many causes (clusters) are projected on top of one another (overlap one another) in the symptom space. The way that a data scientist resolves that degeneracy (another data science word) is to introduce more parameters (higher variety data) in order to “look at” those overlapping clusters from different angles and perspectives, thus resolving the different diagnosis clusters. High variety data enables the discovery of multiple clusters, and eventually identifies the correct cluster (correct diagnosis, in this case).

Higher variety data means that we are adding data from other sensors, other signals, other sources, and of different types. Going back to our low-flying airplane example, this has the following application: I not only heard the aircraft (sound = audio data), but I also looked at it (sight = visual data) and I observed its flight path (dynamic change over time = time series data). The proof of my inference about the airplane was in the data! Additional data sources provided the variety of data signals that were needed in order to derive a correct conclusion.

Similarly, when you go to the doctor with that headache, the doctor will start asking about other symptoms (e.g., lack of appetite; or other pains) and may order other medical tests (blood pressure checks, or other lab results). Those additional data sources and sensors provide the variety of data signals that are needed in order to derive the correct diagnosis.

These examples (low-flying aircraft, and headache pain) are representative analogies of a large number of different use cases in every organization, every business, and every process. The more data you have, the better you are able to detect and discover interesting and important phenomena and events. However, the more variety of data you have, the better you are able to correctly diagnose, interpret, understand, gain insights from, and take appropriate action in response to those phenomena and events.

High-variety data is the fuel that powers these insights, because variety is definitely the secret sauce for bigger and better discovery from big data collections.

Follow Kirk on Twitter at @KirkDBorne

Are Your Predictive Models like Broken Clocks?

A wise philosopher (or comedian) once said, “Even a broken clock is right twice a day.” That same statement might also apply to some predictive models. Since prediction is about the future (usually), then random chance (like broken clockwork) may allow our model to be right occasionally (just by accident). The important step in the data science process that aims to reduce the danger of this occurring is the all-important cross-validation phase (or model-testing phase, which uses an independent data set). This phase is devoted to validating that our model works accurately on previously unseen data that were not used in the model training (model-building) phase.

Another way of characterizing this phase can be found in the field of System Engineering: V&V (Verification and Validation). In the first phase (verification), we verify that the system was built correctly (according to a set of requirements and specifications). In the second phase (validation), we validate that we built the correct system (consistent with the operational needs that the end-user, customer, or client expects the system to satisfy). We sometimes say it this way: (1) in verification, we ask “Did we build the system right?”; and (2) in validation, we ask “Did we build the right system?”

Applying the V&V system engineering principle to data science means that we see model-testing as a two-step process. First, we verify that the model is a logical consequent of the input data used to train the model. Second, we validate that the model remains useful, accurate, and robust when applied to previously unseen data. Any data scientist who participates in Kaggle competitions understands and “lives” this process. It is often the case that our first data science model will do a great job on the “seen” data set (i.e., verification by using a “broken clock” that is right on occasion), but the model then performs poorly on the “unseen” data set. A model that does well on both data sets is a winning model (maybe not in every Kaggle competition, but certainly in real-world usage).

Using the same data set both to validate a model and to train the model would be the data science equivalent of “circular reasoning“. This will often lead to “overfitting“, where the initial model is incorrectly trained to reproduce every variation, bump, wiggle, nuance, and noisy deviation in the training data set, thus falsely exaggerating the importance of those fluctuations. “Complexity” describes our world, but it shouldn’t describe our models.

The other extreme in model-building can be just as bad: underfitting (or bias) introduced by using too few explanatory variables to model the behaviors seen in our data set. I like to believe that Albert Einstein understood data science modeling very well when he said “Everything should be made as simple as possible, but not simpler.” Building an excessively complex model (with too many parameters that follow the noise fluctuations in our data) is like putting too much confidence in a broken clock (“it’s exactly right… some of the time!”). George Box warned us to have a little humility in the face of complex data (and a complex world): “All models are wrong, but some are useful.”

Therefore, when faced with highly complex (high-variety) big data, we are also faced with how to choose the “right model”. We should apply the “Goldilocks principle” — choose a model that is not “too good” and not too bad (i.e., the model works well enough on the training data set and on the test data set).

Follow Kirk Borne on Twitter @KirkDBorne

model_complexity_error_training_test

 

(Source for graphic: http://gerardnico.com/wiki/data_mining/bias_trade-off)

Definitive Guides to Data Science and Analytics Things

The Definitive Guide to anything should be a helpful, informative road map to that topic, including visualizations, lessons learned, best practices, application areas, success stories, suggested reading, and more.  I don’t know if all such “definitive guides” can meet all of those qualifications, but here are some that do a good job:

  1. The Field Guide to Data Science (big data analytics by Booz Allen Hamilton)
  2. The Data Science Capability Handbook (big data analytics by Booz Allen Hamilton)
  3. The Definitive Guide to Becoming a Data Scientist (big data analytics)
  4. The Definitive Guide to Data Science – The Data Science Handbook (analytics)
  5. The Definitive Guide to doing Data Science for Social Good (big data analytics, data4good)
  6. The Definitive Q&A Guide for Aspiring Data Scientists (big data analytics, data science)
  7. The Definitive Guide to Data Literacy for all (analytics, data science)
  8. The Data Analytics Handbook Series (big data, data science, data literacy by Leada)
  9. The Big Analytics Book (big data, data science)
  10. The Definitive Guide to Big Data (analytics, data science)
  11. The Definitive Guide to the Data Lake (big data analytics by MapR)
  12. The Definitive Guide to Business Intelligence (big data, business analytics)
  13. The Definitive Guide to Natural Language Processing (text analytics, data science)
  14. A Gentle Guide to Machine Learning (analytics, data science)
  15. Building Machine Learning Systems with Python (a non-definitive guide) (data analytics)
  16. The Definitive Guide to Data Journalism (journalism analytics, data storytelling)
  17. The Definitive “Getting Started with Apache Spark” ebook (big data analytics by MapR)
  18. The Definitive Guide to Getting Started with Apache Spark (big data analytics, data science)
  19. The Definitive Guide to Hadoop (big data analytics)
  20. The Definitive Guide to the Internet of Things for Business (IoT, big data analytics)
  21. The Definitive Guide to Retail Analytics (customer analytics, digital marketing)
  22. The Definitive Guide to Personalization Maturity in Digital Marketing Analytics (by SYNTASA)
  23. The Definitive Guide to Nonprofit Analytics (business intelligence, data mining, big data)
  24. The Definitive Guide to Marketing Metrics & Analytics
  25. The Definitive Guide to Campaign Tagging in Google Analytics (marketing, SEO)
  26. The Definitive Guide to Channels in Google Analytics (SEO)
  27. A Definitive Roadmap to the Future of Analytics (marketing, machine learning)
  28. The Definitive Guide to Data-Driven Attribution (digital marketing, customer analytics)
  29. The Definitive Guide to Content Curation (content-based marketing, SEO analytics)
  30. The Definitive Guide to Collecting and Storing Social Profile Data (social big data analytics)
  31. The Definitive Guide to Data-Driven API Testing (analytics automation, analytics-as-a-service)
  32. The Definitive Guide to the World’s Biggest Data Breaches (visual analytics, privacy analytics)

Follow Kirk Borne on Twitter @KirkDBorne

4_book_image-6fd6043b69f0bb051f45055c9481cccc

Reach Analytics Maturity through Fast Automatic Modeling

The late great baseball legend Yogi Berra was credited with saying this gem: “The future ain’t what it used to be.” In the context of big data analytics, I am now inclined to believe that Yogi was very insightful — his statement is an excellent description of Prescriptive Analytics.

Prescriptive Analytics goes beyond Descriptive and Predictive Analytics in the maturity framework of analytics. “Descriptive” analytics delivers hindsight (telling you what did happen, by generating reports from your databases), and “predictive” delivers foresight (telling you what will happen, through machine learning algorithms). Going one better, “prescriptive” delivers insight: discovering so much about your application domain (from your collection of big data and information resources, through data science and predictive models) that you are now able to take the actions (e.g., set the conditions and parameters) needed to achieve a prescribed (better, optimal, desired) outcome.

So, if predictive analytics can use historical training data sets to tell us what will happen in the future (e.g., which products a customer will buy; where and when your supply chain will need replenishing; which vehicles in your corporate fleet will need repairs; which machines in your manufacturing plant will need maintenance; or which servers in your data center will fail), then prescriptive analytics can alter that future (i.e., the future ain’t what it used to be).

When dealing with large high-variety data sets, with many features and measured attributes, it is often difficult to build accurate models that are generally useful under a variety of conditions and that capture all of the complexities of the response functions and explanatory variables within your business application. In such cases, fast automatic modeling tools are needed. These tools can help to identify the minimum viable feature set for accurate predictive and prescriptive modeling. For this purpose, I recommend that you check out the analytics solutions from the fast automatic modeling folks at http://soft10ware.com/.

The Soft10 software package is trained to observe quickly and report automatically the most significant, informative and explanatory dependencies in your data. Those capabilities are the “secret sauce” in insightful prescriptive analytics, and they coincide nicely with another insightful quote from Yogi Berra: “You can observe a lot by just watching.”

(Read the full blog at: https://www.linkedin.com/pulse/prescriptive-analytics-future-aint-what-used-kirk-borne)

Predictive versus Prescriptive Analytics

Predictive Analytics (given X, find Y) vs. Prescriptive Analytics (given Y, find X)

Follow Kirk Borne on Twitter @KirkDBorne

Fraud Analytics: Fast Automatic Modeling for Customer Loyalty Programs

It doesn’t take a rocket scientist to understand the deep and dark connection between big money and big fraud. One need only look at black markets for drugs and other controlled and/or precious commodities. But what about cases where the commodity is soft, intangible, and practically virtual? I am talking about loyalty and rewards programs.

A study by Colloquy (in 2011) estimated that the loyalty and rewards programs in the U.S. alone had an estimated outstanding value of $48 billion US dollars. This is “outstanding” value because it doesn’t carry tangible benefit until the rewards or loyalty points are cashed in, redeemed, or otherwise exchanged for something that you can “take to the bank”. In anybody’s book, $48 billion is really big value — i.e., big money rewards for loyal customers, and a big target for criminals seeking to defraud the rightful beneficiaries of these rewards.

The risk vs. reward equation in loyalty programs now has huge numbers on both sides of that equation. There’s great value for customers. There’s great return on investment for businesses seeking loyal customers. And that’s great bait to lure criminals into the game.

In the modern digital marketplace, it is now possible to manipulate payment systems on a larger scale, thereby defrauding the business of thousands of dollars in rewards points. The scale of the fraud could match the scale of the entire loyalty program for some firms, which would therefore bankrupt their supply of rewards for their loyal and faithful customers. This is a really big problem waiting to happen unless something is done about it.

The something that can be done about it is to take advantage of the fast predictive modeling capabilities for fraud detection that are enabled by access to more data (big data), better technology (analytics tools), and more insightful predictive and prescriptive algorithms (data science).

Fraud analytics is no silver bullet. It won’t rid the world of fraudsters and other criminals. But at least fast automatic modeling will give firms better defenses, more timely alerts, and faster response capabilities. This is essential because, in the digital era, it is not only business that is moving at the speed of light, but so also are the business disruptors.

Some simple use cases for fraud analytics within the context of customer loyalty reward programs can be found in the article “Where There’s Big Money, There’s Big Fraud (Analytics)“.

Payment fraud reaches across a vast array of industries: insurance (of all kinds), underwriting, social programs, purchasing and procurement, and now loyalty and rewards programs. Be prepared. Check out the analytics solutions from the fast automatic modeling folks at http://soft10ware.com/.

Follow Kirk Borne on Twitter @KirkDBorne

 

Open Data: Big Benefits, 7 V’s, and Thousands of Repositories

Open data repositories are fantastic for many reasons, including: (1) they provide a source of insight and transparency into the domains and organizations that are represented by the data sets; (2) they enable value creation across a variety of domains, using the data as the “fuel” for innovation, government transformation, new ideas, and new businesses; (3) they offer a rich variety of data sets for data scientists to sharpen their data mining, knowledge discovery, and machine learning modeling skills; (4) they allow many more eyes to look at the data and thereby to see things that might have been missed by the creators and original users of the data; and (5) they enable numerous “data for social good” activities (hackathons, citizen-focused innovations, public development efforts, and more).

Some of the key players in efforts that use open data for social good include: DataKind, Bayes ImpactBooz-Allen Hamilton, Kaggle, Data Analysts for Social Good, and the Tableau Foundation. Check out this “Definitive Guide to do Data Science for Good.” Interested scientists should also check out the Data Science for Social Good Fellowship Program.

We have discussed 6 V’s of Open Data at the DATA Act Forum in July 2015.  We have now added more. The following seven V’s represent characteristics and challenges of open data:

  1. Validity:  data quality, proper documentation, and data usefulness are always an imperative, but it is even more critical to pay attention to these data validity concerns when your organization’s data are exposed to scrutiny and inspection by others.
  2. Value:  new ideas, new businesses, and innovations can arise from the insights and trends that are found in open data, thereby creating new value both internal and external to the organization.
  3. Variety:  the number of data types, formats, and schema are as varied as the number of organizations who collect data. Exposing this enormous variety to the world is a scary proposition for any data scientist.
  4. Voice:  your open data becomes the voice of your organization to your stakeholders (including customers, clients, employees, sponsors, and the public).
  5. Vocabulary:  the semantics and schema (data models) that describe your data are more critical than ever when you provide the data for others to use. Search, discovery, and proper reuse of data all require good metadata, descriptions, and data modeling.
  6. Vulnerability:  the frequency of data theft and hacking incidents has increased dramatically in recent years — and this is for data that are well protected. The likelihood that your data will be compromised is even greater when the data are released “into the wild”. Open data are therefore much more vulnerable to misuse, abuse, manipulation, or alteration.
  7. proVenance (okay, this is a “V” in the middle, but provenance is absolutely central to data curation and validity, especially for Open Data):  maintaining a formal permanent record of the lineage of open data is essential for its proper use and understanding. Provenance includes ownership, origin, chain of custody, transformations that been made to it, processing that has been applied to it (including which versions of processing software were used), the data’s uses and their context, and more.

Here are some sources and meta-sources of open data:

We have not even tried to list here the thousands of open data sources in specific disciplines, such as the sciences, including astronomy, medicine, climate, chemistry, materials science, and much more.

The Sunlight Foundation has published an impressively detailed list of 30+ Open Data Policy Guidelines at http://sunlightfoundation.com/opendataguidelines/. These guidelines cover the following topics (and more) with several real policy examples provided for each: (a) What data should be public? (b) How to make data public? (c) Creating permanent and lasting access to data. (d) Mandating the use of unique identifiers. (e) Creating public APIs for accessing information. (f) Creating processes to ensure data quality.

Related to open data initiatives, the W3C Working Group for “Data on the Web Best Practices” has published a Data Quality Vocabulary (to express the data’s quality), including the following 10 quality metrics for data on the web (which are related to our 7 V’s of open data that we described above):

  1. Statistics
  2. Availability
  3. Processability
  4. Accuracy
  5. Consistency
  6. Relevance
  7. Completeness
  8. Conformance
  9. Credibility
  10. Timeliness

Follow Kirk Borne on Twitter @KirkDBorne

 

Blogging My Way Through Data Science, Big Data, and Analytics

I frequently write blog posts on other sites.  You can find those articles here (updated March 21, 2016):

I also write “one-off” blog posts, such as these examples:

Follow Kirk Borne on Twitter @KirkDBorne