Knowledge Organization: Make Semantics explicit

The organization of knowledge on the basis of semantic knowledge models is a prerequisite for an efficient knowledge exchange. A well-known counter-example are individual folder systems or mind maps for the organization of files. This approach to knowledge organization only works at the individual level and is not scalable because it is full of implicit semantics that can only be understood by the author himself.

To organize knowledge well, we should therefore use established knowledge organization systems (KOS) to model the underlying semantic structure of a domain. Many of these methods have been developed by librarians to classify and catalog their collections, and this area has seen massive changes due to the spread of the Internet and other network technologies, leading to the convergence of classical methods of library science and from the web community.

When we talk about KOSs today, we primarily mean Networked Knowledge Organization Systems (NKOS). NKOS are systems of knowledge organization such as glossaries, authority files, taxonomies, thesauri and ontologies. These support the description, validation and retrieval of various data and information within organizations and beyond their boundaries.

Let’s take a closer look: Which KOS is best for which scenario? KOS differ mainly in their ability to express different types of knowledge building blocks. Here is a list of these building blocks and the corresponding KOS.

The Simple Knowledge Organization System (SKOS), a widely used standard specified by the World Wide Web Consortium (W3C), combines numerous knowledge building blocks under one roof. Using SKOS, all knowledge from lines 1–4 can be expressed and linked to facts based on other ontologies.

Knowledge organization systems make the meaning of data or documents, i.e., their semantics, explicit and thus accessible, machine-readable and transferable. This is not the case when someone places files on their desktop computer in a folder called “Photos-CheeseCake-January-4711” or uses tags like “CheeseCake4711” to classify digital assets. Instead of developing and applying only personal, i.e., implicit semantics, that may still be understandable to the author, NKOS and ontologies take a systemic approach to knowledge organization.

Basic Principles of Semantic Knowledge Modeling

Semantic knowledge modeling is similar to the way people tend to construct their own models of the world. Every person, not just subject matter experts, organizes information according to these ten fundamental principles:

1. Draw a distinction between all kinds of things: ‘This thing is not that thing.’
2. Give things names: ‘This thing is a cheese called Emmental’ (some might call it Emmentaler or Swiss cheese, but it’s still the same thing).
3. Create facts and relate things to each other: ‘Emmental is made with cow’s milk’, Cow’s milk is obtained from cows’, etc.
4. Classify things: ‘This thing is a cheese, not a ham.’
5. Create general facts and relate classes to each other: ‘Cheese is made from milk.’
6. Use various languages for this; e.g., the above-mentioned fact in German is ‘Emmentaler wird aus Kuhmilch hergestellt’ (remember: the thing called ‘Kuhmilch’ is the same thing as the thing called ‘cow’s milk’—it’s just that the name or label for this thing that is different in different languages).
7. Putting things into different contexts: this mechanism, called “framing” in the social sciences, helps to focus on the facts that are important in a particular situation or aspect. For example, as a nutritional scientist, you are more interested in facts about Emmental cheese compared to, for example, what a caterer would like to know. (With named graphs you can represent this additional context information and add another dimensionality to your knowledge graph.)
8. If things with different URIs from the same graph are actually one and the same thing, merging them into one thing while keeping all triples is usually the best option. The URI of the deprecated thing must remain permanently in the system and from then on point to the URI of the newly merged thing.
9. If things with different URIs contained in different (named) graphs actually seem to be one and the same thing, mapping (instead of merging) between these two things is usually the best option.
10. Inferencing: generate new relationships (new facts) based on reasoning over existing triples (known facts).

Many of these steps are supported by software tools. Steps 7–10 in particular do not have to be processed manually by knowledge engineers, but are processed automatically in the background. As we will see, other tasks can also be partially automated, but it will by no means be possible to generate knowledge graphs fully automatically. If a provider claims to be able to do so, no knowledge graph will be generated, but a simpler model will be calculated, such as a co-occurrence network.

Read more: The Knowledge Graph Cookbook

Source Prolead brokers usa

Big Data in the Healthcare Industry: Definition, Implementation, Risks

How extensive must data sets be to be considered as big data? For some, a slightly larger Excel spreadsheet is “big data”. Fortunately, there are certain characteristics that allow us to describe big data pretty well.

According to IBM, 90% of the data that exists worldwide today was created in the last 2 years alone. Big Data Analysis in Healthcare could be helpful in many ways. For example, such analyzes may also counteract the spread of diseases and optimize the needs-based supply of medicinal products and medical devices.

In this article, we will define what is Big Data and discuss ways it could be applied in Healthcare.

The easiest way to say is: Big data is data that can no longer only be processed by one computer. They are so big that you have to store and edit them piece by piece on several servers.

A short definition can also be expressed by three Vs:

1. Volume – describes the size of the data
2. Variety – a variety of data
3. Velocity – the speed of the data

Volume – The Size of Data

As I said before, big data is most easily described by its sheer volume and complexity. These properties do not allow big data to be stored or processed on just one computer. For this reason, this data is stored and processed in specially developed software ecosystems, such as Hadoop.

Variety – Data Diversity

Mass data is very diverse and can also be structured, unstructured or semi-structured.

These data also mostly have different sources. For example, a bank could store transfer data from its customers, but also recordings of telephone conversations made by its customer support staff.

In principle, it makes sense to save data in the format in which it was recorded. The Hadoop Framework enables companies to do just that: the data is saved in the format in which it was recorded.

With Hadoop, there is no need to convert customer call data into text files. They can be saved directly as audio calls. However, the use of conventional database structures is then also not possible.

Velocity – The Speed of Data

This is about the speed at which the data is saved.

It is often necessary that data be stored in real-time. For companies like Zalando or Netflix, it is thus possible to offer their customers product recommendations in real-time.

There are three most obvious, but fundamentally revolutionizing ways of Big Data usage coupled with artificial intelligence.

1. On the one hand, the monitoring. Significant deviations in essential body data will be automatically enhanced in the future: Is the increased pulse a normal sequence of the staircase just climbed? Or does he point to cardiovascular disease in combination with other data and history? Thus, diseases can be detected in their early stages and treated effectively.

1. Diagnosis is the second one. Where it depends almost exclusively on the knowledge and the analysis capacity of the doctor, whether, for example, the cancer metastasis on the X-ray image is recognized as such, the doctor will use artificially intelligent systems, which become a little smarter with each analyzed X-ray image because of Big Data technology. The error probability in the diagnosis decreases, the accuracy in the subsequent treatment increases.

1. And third, after all, Big Data and artificial intelligence have the potential to make the search for new medicines and other treatment methods much more efficient. Today, countless molecular combinations must first be tested in the Petri dish, then in the animal experiment, and finally in clinical trials on their effectiveness, maybe a new drug in the end. A billion company roulette game, in which the winning opportunities can be significantly increased by computer-aided forecasting procedures, which in turn access a never-existed wealth of research data.

As with every innovation in the health system, it’s about the hopes of people to a longer and healthier life. For the urgent that you could be torn from life prematurely through cancer, heart attack, stroke, or another insidious disease from life.

If you want to examine the case of Big Data in practice, you can check this Big Data in the Healthcare Industry article.

Apache Hadoop Framework

To meet these special properties and requirements of big data, the Hadoop framework was designed as open-source. It basically consists of two components:

HDFS

First: It stores data on several servers (in clusters) as so-called HDFS (Hadoop Distributed File System). Second: it processes this data directly on the servers without downloading it to a computer. The Hadoop system processes the data where it is stored. This is done using a program called MapReduce.

MapReduce

MapReduce processes the data in parallel on the servers, in two steps: first, smaller programs, so-called “mappers”, are used. Mappers sort the data according to categories. In the second step, so-called “reducers” process the categorized data and calculate the results.

Hive

The operation of MapReduce requires programming knowledge. To make this requirement a little easier, another superstructure was created on the Hadoop framework – Hive. Hive does not require any programming knowledge and is based on the HDFS and MapReduce framework.  The commands in Hive are reminiscent of the commands in SQL, a standard language for database applications, and are only then translated in MapReduce in the second step.

The disadvantage: it takes a little more time because the code is still translated into MapReduce.

The amount of data available is increasing exponentially. At the same time, the costs of saving and storing this data also decrease. This leads many companies to save data as a precaution and check how it can be used in the future.  As far as personal data is concerned, there are of course data protection issues.

 In this article, I don’t mean to call a Big Data groundbreaking shot today. I believe it’s something that should be adopted widely, and that already has been taken by a lot of world-famous companies.

In the course of the digitization of the health system in general and currently, also with the corona crisis in particular, there are also new questions for data protection. The development and use of ever further technologies, applications, and means of communication offer a lot of benefits but also carries (data protection) risks. Medical examinations in video chat, telemedicine, attests over the internet and a large number of different health apps mean that health data does not simply remain within an institution like a hospital, but on private devices, on servers of app developers, or other places.

 Firstly we have to deal with the question of which data sets are actually decisive for the question that we want to answer with the help of data analysis. Without this understanding, big data is nothing more than a great fog that obscures a clear view through technology-based security.

Source Prolead brokers usa

Unleashing the Business Value of Technology Part 2: Connecting to Value

Figure 1: Unleashing the Value of Technology Roadmap

In part 1 of the blog series “Unleashing the Business Value of Technology Part 1: Framing the Cha…”, I introduced the 3 stages of “unleashing business value”:  1) Connecting to Value, 2) Envisioning Value and 3) Delivering Value (see Figure 1).

We discussed why technology vendors suck at unleashing the business value of their customers’ technology investments because they approach the challenge with the wrong intent.  We then discussed some failed technology vendor engagement approaches; product-centric approaches that force the customer to “make the leap of faith” across the chasm of value.

We also introduced the Value Engineering Framework as a way to “reframe the customer discussion and engagement approach; a co-creation framework that puts customer value realization at the center of the customer engagement” (see Figure 2).

Figure 2: Value Engineering Framework

The Value Engineering Framework is successful not only because it starts the co-creation relationship around understanding and maximizing the sources of customer value creation, but the entire process puts your customer value creation at the center of the relationship.

In Part 2, we are going to provide some techniques that enable technology vendors to connect to “value”, but that is “value” as defined by the customer, not “value” as defined by product or services capabilities.  The Value Engineering Framework helps transition the customer engagement discussion away from technology outputs towards meaningful and material customer business outcomes (see Figure 3).

Figure 3: Modern Data Governance:  From Technology Outputs to Business Outcomes

So, how do technology vendors “connect to value” in their conversations with customers’ business executives?  They must invest the upfront time to understand where and how value is created by the customer.  Here are a couple of tools and techniques that technology vendors can use to understand and connect with the customer’s sources of value creation.

I’m always surprised by how few technology vendors take the time to read their customers’ financial statements to learn what’s important to their customers. Financial reports, press releases, quarterly analyst calls, corporate blogs and analyst websites (like SeekingAlpha.com) are a rich source of information about an organization’s strategic business initiatives – those efforts by your customers to create new sources of business and operational value.

But before we dive into an annual report exercise, let’s establish some important definitions to ensure that we are talking about the same things:

• Charter or Mission is why an organization exists. For example, the mission for The Walt Disney Company is to be “one of the world’s leading producers and providers of entertainment.”
• Business Objectives describe what an organization expects to accomplish over the next 2 to 5 years. The Business Objectives for The Disney Company might include MagicBand introduction, launch “Black Widow” movie, and launch the new Disney World “Star Wars – Rise of the Resistance Trackless Dark Ride” attraction.
• Business Initiative is a cross-functional effort typically 9-12 months in duration, with well-defined business metrics, that supports the entity’s business objectives. For The Disney Company example, it might be to “leverage the MagicBand to increase guest satisfaction by 15%” or “leverage the MagicBand increase cross-selling of Class B attractions by 10%.”
• Decisions are defined as a conclusion or resolution reached after consideration or analysis that leads to action. Decisions address what to do, when to do it, who does it and where to do it.  For The Disney Company example, “Offer FastPass+ to these guests for these attractions at this time of the day” is an example of a decision.
• Use Cases are a cluster of Decisions around a common subject area in support of the targeted business initiative. The Disney Company use cases supporting the “Increase Class B Attraction Attendance” Business Initiative could include:
  • Increase Class A to Class B attraction cross-promotional effectiveness by X%
  • Optimize Class B attraction utilization using FastPass+ by X%
  • Increase targeted guest park experience using FastPass+ by X%
  • Optimize FastPass+ promotional effectiveness by time-of-day by X%

Using these definitions, let’s examine the Starbucks’ 2019 Annual Report to identify their key business objectives (see Figure 4).

Figure 4: Reading the 2019 Starbucks Annual Report

From Figure 4, we can see that one of Starbucks business objectives is “Providing each customer a unique Starbucks experience.”  (Note:  the annual report is chockfull of great opportunities for technology vendors to co-create value with their customers). Let’s triage Starbuck’s “Unique Starbucks Experience” business objective to understand our technology product and service capabilities can enable their “Providing each customer a unique Starbucks experience”.   Welcome to the “Big Data Strategy Document”.

The Big Data Strategy Document decomposes an organization’s business objective into its potential business initiatives, desired business outcomes, critical success factors against which progress and success will be measured, and key tasks or actions. The Big Data Strategy Document provides a design template for contemplating and brainstorming the areas where the technology vendor can connect to the customer’s sources of value creation prior to ever talking to a customer’s business executives. This includes the following:

1. Business Objective. The title of the document states the 2 to 3-year business strategy upon which big data is focused.
2. Business Initiatives. This section states the 9 to 12-month business initiatives that supports the business strategy (sell more memberships, sell more products, acquire more new customers).
3. Desired Outcomes. This section contains the Desired Business or Operational Outcomes with respect to what success looks like (retained more customers, improved operational uptime, reduced inventory costs).
4. Critical Success Factors (CSF). Critical Success Factors list the key capabilities necessary to support the Desired Outcomes.
5. Use Cases. This section provides the next level of detail regarding the specific use cases (“how to do it”) around which the different part of the organizations will need to collaborate to achieve the business initiatives.
6. Data Sources. Finally, the document highlights the key data sources required to support this business strategy and they key business initiatives.

(Note:  the Big Data Strategy Document is covered in Chapter 3 my first book “Big Data: Understanding How Data Powers Big Business.”  The book provides worksheets to help organizations to determine where and how big data can derive and drive new sources of business and operational value.  Still a damn relevant book!)

See the results of the Starbucks triage exercise in Figure 5.

Figure 5:  Starbucks Big Data Strategy Document

To learn more about leveraging the Big Data Strategy Document, check out this oldie but goodie blog “Most Excellent Big Data Strategy Document”.

The challenge most technology vendors face when trying to help their customers unleash the business value of their technology investments, is that vendors don’t intimately understand how their customers create value.  Once the technology vendor understands how the customer creates value, then the technology vendor has a frame against which to position their product and service capabilities to co-create new sources of value for both the customer and the technology vendor.

Source Prolead brokers usa

Defining and Measuring Chaos in Data Sets: Why and How, in Simple Words

There are many ways chaos is defined, each scientific field and each expert having its own definitions. We share here a few of the most common metrics used to quantify the level of chaos in univariate time series or data sets. We also introduce a new, simple definition based on metrics that are familiar to everyone. Generally speaking, chaos represents how predictable a system is, be it the weather, stock prices, economic time series, medical or biological indicators, earthquakes, or anything that has some level of randomness. 

In most applications, various statistical models (or data-driven, model-free techniques) are used to make predictions. Model selection and comparison can be based on testing various models, each one with its own level of chaos. Sometimes, time series do not have an auto-correlation function due to the high level of variability in the observations: for instance, the theoretical variance of the model is infinite. An example is provided in section 2.2 in this article  (see picture below), used to model extreme events. In this case, chaos is a handy metric, and it allows you to build and use models that are otherwise ignored or unknown by practitioners.  

Figure 1: Time series with indefinite autocorrelation; instead, chaos is used to measure predictability

Below are various definitions of chaos, depending on the context they are used for. References about how to compute these metrics, are provided in each case.

Hurst exponent

The Hurst exponent H is used to measure the level of smoothness in time series, and in particular, the level of long-term memory. H takes on values between 0 and 1, with H = 1/2 corresponding to the Brownian motion, and H = 0 corresponding to pure white noise. Higher values correspond to smoother time series, and lower values to more rugged data. Examples of time series with various values of H are found in this article, see picture below. In the same article, the relation to the detrending moving average (another metric to measure chaos) is explained. Also, H is related to the fractal dimension. Applications include stock price modeling.

Figure 2: Time series with H = 1/2 (top), and H close to 1 (bottom)

Lyapunov exponent

In dynamical systems, the Lyapunov exponent is used to quantify how a system is sensitive to initial conditions. Intuitively, the more sensitive to initial conditions, the more chaotic the system is. For instance, the system xn+1 = xn – INT(xn), where INT represents the integer function, is very sensitive to the initial condition x0. A very small change in the value of x0 results in values of xn that are totally different even for n as low as 45. See how to compute the Lyapunov exponent, here.

Fractal dimension

A one-dimensional curve can be defined parametrically by a system of two equations. For instance x(t) = sin(t), y(t) = cos(t) represents a circle of radius 1, centered at the origin. Typically, t is referred to as the time, and the curve itself is called an orbit. In some cases, as t increases, the orbit fills more and more space in the plane. In some cases, it will fill a dense area, to the point that it seems to be an object with a dimension strictly between 1 and 2. An example is provided in section 2 in this article, and pictured below. A formal definition of fractal dimension can be found here.

Figure 3: Example of a curve filling a dense area (fractal dimension  >  1)

The picture in figure 3 is related to the Riemann hypothesis. Any meteorologist who sees the connection to hurricanes and their eye, could bring some light about how to solve this infamous mathematical conjecture, based on the physical laws governing hurricanes. Conversely, this picture (and the underlying mathematics) could also be used as statistical model for hurricane modeling and forecasting. 

Approximate entropy

In statistics, the approximate entropy is a  metric used to quantify regularity and predictability in time series fluctuations. Applications include medical data, finance, physiology, human factors engineering, and climate sciences. See the Wikipedia entry, here.

It should not be confused with entropy, which measures the amount of information attached to a specific probability distribution (with the uniform distribution on [0, 1] achieving maximum entropy among all continuous distributions on [0, 1], and the normal distribution achieving maximum entropy among all continuous distributions defined on the real line, with a specific variance). Entropy is used to compare the efficiency of various encryption systems, and has been used in feature selection strategies in machine learning, see here.

Independence metric 

Here I discuss some metrics that are of interest in the context of dynamical systems, offering an alternative to the Lyapunov exponent to measure chaos. While the Lyapunov exponents deals with sensitivity to initial conditions, the classic statistics mentioned here deals with measuring predictability for a single instance (observed time series) of a dynamical systems. However, they are most useful to compare the level of chaos between two different dynamical systems with similar properties. A dynamical system is a sequence xn+1 = T(xn), with initial condition x0. Examples are provided in my last two articles, here and here. See also here. 

A natural metric to measure chaos is the maximum autocorrelation in absolute value, between the sequence (xn), and the shifted sequences (xn+k), for k = 1, 2, and so on. Its value is maximum and equal to 1 in case of periodicity, and minimum and equal to 0 for the most chaotic cases. However, some sequences attached to dynamical systems, such as the digit sequence pictured in Figure 1 in this article, do not have theoretical autocorrelations: these autocorrelations don’t exist because the underlying expectation or variance is infinite or does not exist. A possible solution with positive sequences is to compute the autocorrelations on yn = log(xn) rather than on the xn‘s.

In addition, there may be strong non-linear dependencies, and thus high predictability for a sequence (xn), even if autocorrelations are zero. Thus the desire to build a better metric. In my next article, I will introduce a metric measuring the level of independence, as a proxy to quantifying chaos. It will be similar in some ways to the Kolmogorov-Smirnov metric used to test independence and illustrated here, however, without much theory, essentially using a machine learning approach and data-driven, model-free techniques to build confidence intervals and compare the amount of chaos in two dynamical systems: one fully chaotic versus one not fully chaotic. Some of this is discussed here.

I did not include the variance as a metric to measure chaos, as the variance can always be standardized by a change of scale, unless it is infinite.

To receive a weekly digest of our new articles, subscribe to our newsletter, here.

About the author:  Vincent Granville is a data science pioneer, mathematician, book author (Wiley), patent owner, former post-doc at Cambridge University, former VC-funded executive, with 20+ years of corporate experience including CNET, NBC, Visa, Wells Fargo, Microsoft, eBay. Vincent is also self-publisher at DataShaping.com, and founded and co-founded a few start-ups, including one with a successful exit (Data Science Central acquired by Tech Target). You can access Vincent’s articles and books, here.

Source Prolead brokers usa

A simple way for getting started with fast.ai for pytorch

At my AI course in the University of Oxford, we are exploring the use of PyTorch for the first time.

One of the best libraries to get started with PyTorch is fast.ai.

There are various ways to learn fast.ai.

For most people, the fast.ai course is their first exposure.

There is now a book which I recently bought Deep Learning for Coders with fastai and PyTorch By Jeremy Howard and Sylvain Gugger

However, there is also a paper by the creators.

 I found this paper as a concise starting point fastai: A Layered API for Deep Learning

In this post, I use the paper to provide a big picture overview of fast.ai because it helped me to understand the library in this way.

fastai is a modern deep learning library, available from GitHub as open source under the Apache 2 license. The original target of the API was for beginners and also practitioners who are interested in applying pre-existing deep learning methods.  The library offers APIs targeting four application domains: vision, text, tabular and time-series analysis, and collaborative filtering. The idea here is to choose intelligent default values and behaviors for the applications.  

While the high level API is targeted at solution developers, the mid-level API provides the core deep learning and data-processing methods for each of these applications. Finally, the  low-level APIs provide a library of optimized primitives and functional and object-oriented foundations, which allows the mid-level to be developed and customised.

Mid level APIs include functions like Learner, Two-way callbacks, Generic optimizer, Generalized metric API, fastai.data.external, funcs_kwargs and DataLoader, fastai.data.core, Layers and architectures

The low-level of the fastai stack provides a set of abstractions for: Pipelines of transforms,  Type-dispatch,   GPU-optimized computer vision operations etc

Finally, there is a programming environment called nbdev, which allows users to create complete Python packages.  

The Mid level APIs are a key differentiator for fast.ai because it allows more developers to customise the software in contrast to a small community of specialists.

To conclude, the carefully layered design makes fast.ai highly customizable (especially the mid level API) enabling more users to build their own applications or customize the existing ones.

Image source:  fast.ai

Source Prolead brokers usa

Strategies for a successful Voice of the Customer program

It is more important than ever to retain customers. Success often relies on having a deep understanding of your customers across every touch point –and that involves listening. That’s where an effective Voice of the Customer program can add real value, delivering insights to help you improve customer experience and meet key business objectives.

To build and run a successful Voice of the Customer program, your approach will evolve along the way, so think about it in three strategic phases: getting a great start, building momentum, and then expanding the potential.

Phase 1: Plan for success with your Voice of the Customer Program

• Create a Strategic Roadmap: No matter how large or small your organization, or what industry you are in, you’ll gain greater value at lower cost if your Voice of the Customer program starts with a clear game plan.

• Gain a holistic view of the customer experience: To really understand and improve your customer’s experience, it’s important to develop a complete picture of their relationship. For well-rounded insights, be sure to monitor numerous touch points —capturing both structured data (e.g., surveys and transaction data) and unstructured data (e.g., call center transcripts and customer support email feedback). And don’t forget to track social media, where customers often vent about, or praise, their service experiences. Analyzing both structured and unstructured data provides a richer, more nuanced view of the customer experience. Additionally, it’s a good idea to map the customer experience lifecycle (such as pre-sales vs. servicing) to better understand where and how to make improvements.
  Effective Voice of the Customer programs both listen and take action.
• Be prepared to take action to drive improvements: To ensure you can act on insights gained through VOC analytics, build buy-in for customer experience changes by recruiting champions, influencers, and executives across numerous lines of business. To build the business case, start with small, measurable, pilot efforts. As an example, VOC analytics helped a Top 50 bank we worked with uncover numerous customer complaints about being required to make wire transfers in person at banking locations. In response, the bank began offering wire services online, and developed metrics to track the impact of the change.

Phase 2: Optimize your VOC efforts

• Discover more by letting the data speak: You’ll gain more value from your Voice of the Customer program by listening to what customers are really saying. By using natural language processing (NLP) and text analytics to let themes emerge, you unlock the true value of your data. With a more complete picture, you can prioritize targeted improvements that will produce the biggest wins.

• Increase the relevance of insights with unique business context: Your company likely has a wealth of customer comments from surveys, call centers, email and in-store feedback, and social media—so how do you make the most of it? Find out what’s really driving the comments by engaging team members from various lines of business who understand the issues and can provide important context to help classify customer comments. Root-cause analysis can also help you focus on making changes that will mean the most to customers.

• Measure the effectiveness of your actions: To confirm the business value of your Voice of the Customer program, you should consistently track the impact of any improvements you make. Define metrics and leverage analytics dashboards to create progress reports you can share with business leaders across the company. With tools like Domo, Tableau, and Cognos this has gotten easier than ever.

Phase 3: Take your Voice of the Customer program to the next level

• Think bigger by multi-purposing customer insights: Increase the power of your Voice of the Customer program by leveraging insights to make improvements in multiple areas. For example, after analyzing millions of customer comments, you might identify key pain points that enable you to triage customers into different support strategies that help strengthen relationships. Expand your perspective to include feedback from frontline employees and other key partners who play a role in shaping the customer experience. This added layer of insight can help you define strategies for new product offerings, training, or other resources that would appeal to customers and grow your business.

• Increase revenue potential through customer insights: Customer listening can identify more than just the problems; it’s a great way to learn what people value most about your business. From there you can use predictive modeling and machine learning to classify customer segments most likely to respond to certain promotions, and deliver targeted marketing. You can also leverage VOC analytics to “crowd source” for ideas on how to attract more business. In particular, social media analytics may uncover insights about what people want that you don’t already offer.

• Build more power into CRM with insights from VOC: Boost the value of your Customer Relationship Management (CRM) program by systematically tracking feedback as part of your customer profiles. By integrating customer comments from multiple touch points into CRM, you can better understand their emotional connection to your brand. It also helps you identify customers who consistently provide positive feedback so you can explore cross-sell or up-sell opportunities, and even engage them to become brand advocates.

To gain the most from your Voice of the Customer program, focus your approach on advanced analytics. Many companies do a great job of listening and gathering data, but don’t maximize the potential to create customer insights and drive action. Without action, there is no ROI from your listening efforts. When you increase the rigor and maturity of your VOC analytics, you can use what you learn about customers to drive measurable change and improve customer experience.

Our team is passionate about VOC, check out our other blogs about Voice of the Customer. 

​​​​​​​​

Source Prolead brokers usa

On-demand Mobile Apps: Must-have Features and Trends

The success of Uber has changed the approach to business operations in the tertiary sector and inspired hundreds of startups. A smartphone has become a platform for receiving services in a new hassle-free way for users, and the distance between them and the service providers has decreased. The latter also got the possibility to build a robust online presence and connect to the target audience directly. Technology companies are another essential component in this process as they develop customized solutions to cater to consumer needs. Healthcare, education, retail, transportation, beauty, etc. – the list of industries taking advantage of the economy uberization is vast. Even the smallest companies or individuals-freelancers can instantly access the consumer base through third-party apps – deliver home-made food, rent out accommodation, offer childcare, cleaning, consulting services, to name just a few. Besides, clients have a tool to shape the market by assessing the quality of products and services through a rating system.

Source Prolead brokers usa

All the Skills Require For A Data Scientist

Source Prolead brokers usa

Markov Decision Processes

The Markov Decision Process (MDP) provides a mathematical framework for solving the RL problem. Almost all RL problems can be modeled as an MDP. MDPs are widely used for solving various optimization problems. In this section, we will understand what an MDP is and how it is used in RL.

To understand an MDP, first, we need to learn about the Markov property and Markov chain.

The Markov property and Markov chain

The Markov property states that the future depends only on the present and not on the past. The Markov chain, also known as the Markov process, consists of a sequence of states that strictly obey the Markov property; that is, the Markov chain is the probabilistic model that solely depends on the current state to predict the next state and not the previous states, that is, the future is conditionally independent of the past.

For example, if we want to predict the weather and we know that the current state is cloudy, we can predict that the next state could be rainy. We concluded that the next state is likely to be rainy only by considering the current state (cloudy) and not the previous states, which might have been sunny, windy, and so on.

However, the Markov property does not hold for all processes. For instance, throwing a dice (the next state) has no dependency on the previous number that showed up on the dice (the current state).

Moving from one state to another is called a transition, and its probability is called a transition probability. We denote the transition probability by  . It indicates the probability of moving from the state s to the next state . Say we have three states (cloudy, rainy, and windy) in our Markov chain. Then we can represent the probability of transitioning from one state to another using a table called a Markov table, as shown in Table 1.1:

Table 1: An example of a Markov table

From Table 1, we can observe that:

• From the state cloudy, we transition to the state rainy with 70% probability and to the state windy with 30% probability.
• From the state rainy, we transition to the same state rainy with 80% probability and to the state cloudy with 20% probability.
• From the state windy, we transition to the state rainy with 100% probability.

We can also represent this transition information of the Markov chain in the form of a state diagram, as shown in Figure 1:

Figure 1: A state diagram of a Markov chain

We can also formulate the transition probabilities into a matrix called the transition matrix, as shown in Figure 2:

Figure 2: A transition matrix

Thus, to conclude, we can say that the Markov chain or Markov process consists of a set of states along with their transition probabilities.

The Markov Reward Process

The Markov Reward Process (MRP) is an extension of the Markov chain with the reward function. That is, we learned that the Markov chain consists of states and a transition probability. The MRP consists of states, a transition probability, and also a reward function.

A reward function tells us the reward we obtain in each state. For instance, based on our previous weather example, the reward function tells us the reward we obtain in the state cloudy, the reward we obtain in the state windy, and so on. The reward function is usually denoted by R(s).

Thus, the MRP consists of states s, a transition probability  , and a reward function R(s).

The Markov Decision Process

The Markov Decision Process (MDP) is an extension of the MRP with actions. That is, we learned that the MRP consists of states, a transition probability, and a reward function. The MDP consists of states, a transition probability, a reward function, and also actions. We learned that the Markov property states that the next state is dependent only on the current state and is not based on the previous state. Is the Markov property applicable to the RL setting? Yes! In the RL environment, the agent makes decisions only based on the current state and not based on the past states. So, we can model an RL environment as an MDP.

Let’s understand this with an example. Given any environment, we can formulate the environment using an MDP. For instance, let’s consider the same grid world environment we learned earlier. Figure 3 shows the grid world environment, and the goal of the agent is to reach state I from state A without visiting the shaded states:

Figure 3: Grid world environment

An agent makes a decision (action) in the environment only based on the current state the agent is in and not based on the past state. So, we can formulate our environment as an MDP. We learned that the MDP consists of states, actions, transition probabilities, and a reward function. Now, let’s learn how this relates to our RL environment:

States – A set of states present in the environment. Thus, in the grid world environment, we have states A to I.

Actions – A set of actions that our agent can perform in each state. An agent performs an action and moves from one state to another. Thus, in the grid world environment, the set of actions is up, down, left, and right.

Transition probability – The transition probability is denoted by . It implies the probability of moving from a state  to the next state  while performing an action . If you observe, in the MRP, the transition probability is just  that is, the probability of going from state  to state  and it doesn’t include actions. But in MDP we include the actions, thus the transition probability is denoted by .

For example, in our grid world environment, say, the transition probability of moving from state A to state B while performing an action right is 100% then it can be expressed as: . We can also view this in the state diagram as shown below:

Figure 4: Transition probability of moving right from A to B

Suppose our agent is in state C and the transition probability of moving from state C to state F while performing the action down is 90%, then it can be expressed as P(F|C, down) = 0.9. We can also view this in the state diagram, as shown in Figure 5:

Figure 5: Transition probability of moving down from C to F

Reward function – The reward function is denoted by . It implies the reward our agent obtains while transitioning from a state  to the state  while performing an action .

Say the reward we obtain while transitioning from state A to state B while performing the action right is -1, then it can be expressed as R(A, right, B) = -1. We can also view this in the state diagram, as shown in Figure 6:

Figure 6: Reward of moving right from A to B

Suppose our agent is in state C and say the reward we obtain while transitioning from state C to state F while performing the action down is +1, then it can be expressed as R(C, down, F) = +1. We can also view this in the state diagram, as shown in Figure 7:

Figure 7: Reward of moving down from C to F

Thus, an RL environment can be represented as an MDP with states, actions, transition probability, and the reward function. Learn more in Deep Reinforcement Learning with Python, Second Edition by Sudharsan Ravichandiran.

Source Prolead brokers usa

How Agile Methodology Assists the Development of Smart and Working Software Products?

As with any project, an effective and result-driven is a must, the same goes with software development projects. There’s no secret that software development is one of the trickiest and complex tasks that ever existed and to do them right, one needs to be following certain methodologies and processes that make the project go smooth and results in a functional project.

Agile is one of such methodology that is widely adopted by software development companies across the globe. It was formally launched back in 2001 when 17 different IT experts joined forces together to set up a software development method that eliminates all the factors contributing to slowing down a project’s development phase. They came up with four points in their famous Agile Manifesto that were driven to make the software development process fast and result-oriented. The 4 points are:

• Individuals and interactions over processes and tools
• Working software over comprehensive documentation
• Customer collaboration over contract negotiation
• Responding to change over following a plan

Today, every organization seems to have been practicing agile methods in rendering software development services knowingly or unknowingly. No matter if you are new to the software development world or you have learned software development a decade after implementing old development methodologies, one thing is for sure that you are somewhere influenced by the agile software development methodologies.

Before proceeding further, it is important to discuss what the most important roles are in agile methodology. Agile methodologies are all about visioning such plans and models that are made to facilitate users so agile methodology hierarchy starts with keeping users in mind. Let’s get some important roles listed down here.

Important Roles in Agile Process:

1. Consumers:

As earlier said in agile methodology, strategies regarding the design and development of software are made to address users’ ultimate needs. In the modern-day, this is labeled as estimating user personas of how users approach a specific software application.

There are hundreds and thousands of software applications available for users, keeping your business software product competitive in this tech-dominated spectrum is key. And here’s agile methodology comes into play. It leverages the whole development process through the workflows that facilitate the development process in a way that brings productivity.

2. The Head of the Product:

A software product is someone’s idea that was once cooking in the brains. That person who envisions a software product is known as the head of the product. This person knows how things will dish out in the process of getting the desired product developed.

These are the persons who act as vocals for end users. They do establish their connections with both the users and the developers to convey users’ needs from a specific software product to developers. In short, they are the brains behind any software product idea. They present their vision of software in a series of demonstrations and hence the execution goes underway for both the long and short time. Priority is given to the segments that are most important and require greater effort.

Apart from coming up with a vision of a product, these people are responsible to interact with the design and the development teams and supervise the project managers to eliminate any discrepancy in the process. This is by far the most amazing benefit of agile methodology that it gets the whole team on the board and tasks are then assigned to them. When a product head interacts with the development team, he or she makes sure they come up with the stories helping developers to understand the user persona. These stories are short and taken from the real-life experiences of users.

These user stories are prioritized by the product owner, reviewed by the team to ensure they have a shared understanding of what is being asked of them.

3. The Team Responsible for Software Development:

These are the people that are directly responsible to come up with a product that is envisioned by the product head. In an agile environment, the members of the software development work on specified and sometimes non-specified tasks.

Because the goal in front of them is very simple; to deliver a functional product incorporating all the features that are supposed to have in the ideation phase. They use each other’s core specialties where it is necessary. In agile environments, the development teams collaborate in a way that they use each other’s skills in a unidirectional way to get the job done.

In some cases, the developer team is not limited to just software developers. They can bring in QA analysts, project managers, designers, and other business analysts to broaden the scope of the development process.

Why Agile Methodology Works Better Than Other Methodologies?

When you pile up factors such as agile development, quick adoption, flexibility, using collaborative development tools, and the right teams, the desired results can be obtained quickly and productively. The agile methodology is all about adaption and flexibility that adds to the whole development process and refines it.

Agile architecture is better for many challenges because its concepts, frameworks, and procedures are based on the working conditions of today. Agile structures and development processes that give priority to the delivery of working applications and encourage input to enhance application and processes are best suited to the smarter and faster environment of today’s operation.

The Wrap:

Finally, the agile procedures are most suited and widely accepted among the software development teams across the globe. Many developers have their say on agile development as a method that has brought more productivity into their work despite putting unnecessary burdens on them.

If you are an entrepreneur and are looking to get your software product developed in the latest fashion, agile methodology is the way forward for you. Moreover, SoftCircles is one of the top software development companies in New York that strictly follows an agile methodology to bring the best out of a software development project.

Source Prolead brokers usa