Bringing Predictive Data Analytics to the People with

Don’t Be Left Behind in the Days Where Predictive Analytics is Mandatory
for Long-Term Business Success


Over the last decade, the term “big data” grew to prominence. The rush was on to create technologies to capture and store vast quantities of data. The focus of many enterprises, both large and small, was on data capture and storage. Now, the rush is on to monetize and exploit these sizable data stores. Companies want to make the right decisions, for the right customers at the precise time to maximize value and minimize risks. In short, businesses need to predict the future by anticipating behaviors and identifying trends, at the individual customer level and at scale across their entire customer base.The company that can achieve all of this before their competition will win the in the marketplace. Companies that don’t capitalize on their data resources by converting them to insights and actions will lose market share, fall behind, and, ultimately, fail.

Data: The New Oil

In the early 20th century there were hardly any automobiles. Accordingly, there were no gas stations or car mechanics. Over time, gas stations evolved to have convenience stores attached to them, auto repair shops thrived, and the insurance industry found a solid new foundation for an entirely new line of business.

Look around you and you will see an entire civilization transformed by oil with all its benefits and detriments. Now, imagine what the world will look like one hundred years from now when the data revolution has played out all its effects and unintended consequences.

Pushing the “data is the new oil” analogy further, consider that raw data exists in a natural, unprocessed state, very often deep underground. A considerable amount of labor goes into taking it from that primordial state into something that can be used to fuel a car or heat a home. The data must be extracted, shaped and processed in a process analogous to what oil refineries do. Finally, the output of the refinery gets sold as a product to consumers. In other words, as more oil does not make for better gasoline, more data doesn’t necessarily make your business data-centric.

Turning Raw Data into Insight

Over the last few years, more and more organizations have discovered that data can be turned into any number of Artificial Intelligence (AI), Machine Learning (ML), or other “cognitive” services. Some of these new services may blossom into new revenue streams and will more than likely disrupt entire industries as the normal way of business will be up ended in favor of automation and accelerated decision making.

Collecting raw data for the sake of collecting raw data, argues Hal Varian, Google’s chief economist, exhibits “decreasing returns to scale.” In other words, each additional piece of data is somewhat less valuable and at some point, collecting more does not add anything. What matters more, he says, is the quality of the algorithms that process the data and the talent a firm has brought on to develop these algorithms. Success for Google is in the “recipe” not the “ingredients.”

As for new world of data, the product could be a service that rates the likelihood of whether or not a transaction is fraudulent and the consumers of the service are internal auditing department. In this way, data will enable new markets and even economic ecosystems as a previously undervalued resource develops into new streams of income and creates entirely new offshoot industries.

Drawbacks of Conventional Analytics

With the future of their businesses at stake, one would think that every single enterprise would be eagerly scouring their data sets and feeding them to any number of algorithms in order to extract any deeper understanding of their customers’ activities and identify trends as they unfold. However, this is not the case. Why?

The answer comes down to cost, both in terms of finding the people with the skills to perform this type of work and in the computation infrastructure often required to run existing algorithms, and risk, as it is often said the value is quite difficult to foresee and the complexity difficult to handle along the analytics initiatives.

This is not mere risk aversion or fear of the unknown: there are hurdles everywhere, indeed. Data will need to be shaped and cleaned; the team is not skilled enough, and hiring on consultants is expensive. Not to mention the infrastructure investments required to store the data and compute the model. The payoff is hard to evaluate and the ROI is even harder to envision.

In a few words: getting tangible results from analytics is not straightforward. At least, not for everyone.

What if there was a way to take the cost of recruiting experienced data scientists and data engineers, remove the expense associated with beefing up IT infrastructure, and make advanced data analytics more approachable to the average knowledge worker?

Well, there is.

Enter changes the game by making advanced analytics more accessible and affordable. No longer are advanced analytics limited to large organizations with massive budgets devoted towards hiring, training, and maintaining a data science team. Now, anyone with or without data science and machine learning skills can leverage the power with a few clicks of the mouse.

Simple to Use

The real power of lies in its ability to place data analytics into an easy to use self-service SaaS model. is now available on the AWS Marketplace. Just activate an account on the marketplace and pay as you go. No software installs or commitments.

Automatic, Swift & Agile Integrated Predictive Analytics automates much of the work normally associated with machine learning. Using autoML algorithms, surfaces and evaluates new data features and only displays the ones with meaningful impact on predictive outcomes. In other words, the software automatically filters out the fields, or features, that lack correlation to the predicted outcome. From the meaningful features it discovers, then creates a predictive model for future input. As the workflow and the display are made straightforward and intuitive, users can focus on rapidly iterating data models, exploring the data and, finally, delivering added value to the business.

Blazingly Fast Heterogeneous or Homogeneous Data Integration

Getting data into is fast and easy. Simply drag and drop ASCII or UTF-8 encoded CSV files: once the primary dataset file is uploaded, users can upload any number of additional peripheral data tables. It’s then up to, supervised by the user and its business knowledge, to detect, surface and display meaningful features and insights from those multiple datasets.

Accessible Advanced Data Analytics

Making advanced analytics accessible opens up new worlds of possibilities. With the freedom of self-service analytics, all of different scenarios are possible. Marketing and sales departments can determine the customers most likely to leave for a competitor – before they leave. They can pre-emptively identify which accounts are high priority calls for the sales team. Outbound calls from the sales center can be optimized to increase conversion and sales performance. Marketing and sales teams can be self-sufficient with their model creation, exploration, and experimentation.

However, the advantages go beyond marketing and sales, business analysts can leverage their deep domain knowledge and apply predictive analytics to pre-emptively address challenges that the business faces and take corrective action ahead of time. They can also use the built-in sharing functionality to share insights with their colleagues and management.

Speaking of management, company leadership can make dashboards into foresight, predicting the course of the business. Using the same tools, senior management can take proactive steps to steer the company around dangers, risks, and even find new opportunities that they may have otherwise missed.

Finally, even seasoned data scientists can leverage the flexibility and power of to explore models with ease and speed. Data science teams can explore more datasets in less time, allowing for them to explore more options, create more effective models, and add more value to the business.

In short, allows non-data scientists to reap the rewards of data science and makes data scientists more efficient.

It is pointless and painful not to use!

Getting Started

The truly best way to see how works is to use it and see for yourself just how approachable it makes data analytics.

Now that is available on the AWS Marketplace, it couldn’t be easier to use. No software to download, nothing to install, no new hardware to provision: it’s just a service that runs in the AWS cloud.

The first step to using is to browse over to the AWS Marketplace and search for It will appear as one of the options in the autocomplete dropdown.

Then choose from one of the following options. For the purposes of this blog post, choose the (single-user) option.

On the next page, choose the AWS Region and EC2 Instance Type you wish to use. Then click on the “Launch with 1-click” button.

On the following page, click on the EC2 Console link to browse to the EC2 console.

On the EC2 Console page, retrieve the URL where was deployed. It will be the domain name next to Public DNS and end with “”.

Copy the URL and then browse to it.

You should see the software home page with no projects.

Data Modeling 101

Now you’re about to create your first data model and discover what lies hidden inside the data: the way. You may be asking yourself, exactly what is a data model? A data model is a way to store and represent data and its relation to other data. For instance, customers have various attributes about them stored in the data model and customer actions are also logged in the data model.

By connecting a customer’s attribute, such as age, to their behavior, such as buying certain products, one can infer that customers of a similar age are more likely to make the same purchases. Many relationships are well known, parents with young children are more likely to buy diapers than those whose children are in college.

What’s great about machine learning and data analytics is the ability to identify patterns quickly and even see patterns that exist that humans may not be able to readily identify. How does that work? Well, let’s create a project and see how easy makes this process.

Creating Your First Project

Click on the Create button located on the upper right-hand corner of the screen. The following dialog box asks for a name for the new project. You may enter anything, but, for this example, I entered “first project.”

Click Create and then the “first project” project now appears in the workspace area.

Click on the first project button to bring up the screen below:

Now drag and drop files to add data to the project. As mentioned previously, data files must be in CSV format. Once the appropriate files are uploaded, specify the one containing the outcome.

Once completed, click the Get Insights button in the upper right hand corner of the screen and will get to work analyzing the data.

After a few moments, the results are in and we can now explore the results.

First, click on the magnifying glass icon to view the model. Click around and explore the features of the data. Note how the visualizations change with each field.

To get back to the previous screen, click the drop down list that starts with “first project” and then choose Models List.

Now click on the chart icon to assess the quality of the model

The following page explains how well the predicted outcomes the model created matched up with the test data.

This screen contains a lot of information. However, if you look at the Performance number, the model scored a “0.6941,” meaning it was correct around 69.41% of the time.

Certainly, there is room for improvement and provides ways for you to adjust and improve the model.

Go back to the models list page and this time, click on the lightning bolt icon to improve the model.

This page allows you to manually adjust the features that go into creating the predictive model. Remove the average_basket feature by clicking the checkbox. Now, add the region_code feature by clicking on the check box. Additionally, change the value in the dropdown list in the Type column to Categorical. Your feature set should look like the following.

Click the Apply Changes button in the upper right hand corner to apply the changes to the model and see if the performance has improved.

Upon quick inspection, you can see that the small changes improved the model modestly. Up to 74.11% now. Go back and experiment which fields and options improve the model and which ones have the opposite effect.

Once you’ve made improvements to the model, it’s time to share it.

On the Models List page, click on the magnifying glass icon to view the model. Check all boxes next to the fields you wish to include in the report. Now click on the Get PDF button in the upper right hand corner of the screen. generates a report in a PDF file that you can share.

If you wanted to share the performance metrics of the model, you can do that as well. Go back to the Models List page and this time click on the chart icon to see the Assess Model screen once more. On the upper right hand corner of the screen, there is once again a Get PDF button. Click on it to generate and download a report as PDF file.

Conclusion allows for easy creation and exploration of predictive data models in just a few clicks of the mouse. It gives business users many of the same analytical tools that have previously only been in the hands of data scientists. With wider use and deployment of data analytics, businesses can more easily spot trends, detect fraud, better serve their existing customers, and find new ones.

Data analytics is already changing the game of business and now its power is in your hands. The recommendation is simple: use before your competition does.

This entry was posted in Artificial Intelligence, Data Modeling, Easy AWS AI,, Predictive Analytics and tagged , , , , , , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s