Online education portals like Udacity and Coursera are really changing the world of remote learning in significant ways. By making free and high quality education accessible to a global audience, these platforms are opening up undreamt of possibilities for communities around the world to improve, grow, and prosper in the digital economy of the 21st century. Education at top tier colleges and universities has traditionally been a social and economic privilege, but now anyone can join in the learning revolution by sitting in virtual classrooms with the world’s best and brightest educators. Whether this involves learning how to code and build smart phone apps, or starting up a new business, or learning about public health literacy, the sky is the limit of what’s now possible.

Everything about Web and Network Monitoring

Predictive analytics for the small business – part 5

1.smallIn our last segment we reviewed the process of applying a predictive model to your dataset – going from assessment of software needs to selecting the observation period to selecting the key variables (dependent and independent), choosing a methodology, and then building and testing the model. Predictive modeling is the crux of predictive analytics, where all your data preprocessing efforts are brought to focus on extracting insights to prove your business question.


Once you’ve finished identifying your data, collecting and cleansing it, and building and testing a successful predictive model, it might be easy to assume the project is done. But there is one more crucial aspect of predictive analytics we want to discuss and that concerns the deployment stage.


Deploying Your Predictive Model


It’s one thing to have an elegant model providing you with great predictive power that reveals which customers are most likely to respond to your targeted advertising campaign. It’s entirely another thing to ensure that the model gets used in your organization to drive real business value. Figuring out how the predictive model gets served up to the rest of the company is known as the deployment stage. This can range from a simple spreadsheet with results given to one person to a complex system where multiple data streams must be fed into the model. Setting up these data streams will likely involve additional software engineering work.






In order to turn your data into quality, actionable customer insight your stakeholders will need a way to read and interpret the predictive model. This is where a data visualization or business intelligence tool will help immensely. A popular program for this is Tableau, which allows you to query your data mining results with custom SQL for easy data visualization.


So taken together, we can see that making the output of the predictive model available across the organization is an additional time-consuming, but necessary task. As one source well states, “The deployment of a successful model is sometimes more work than building the model itself.” Given the costs and time committed this far, it’s very important to ensure there is follow through at this crucial juncture.


The Main Takeaways


We opened this series with a look at the way Big Data represents an epic technological shift transforming all levels of business and society. Gartner best summarizes the latest trends as follows: “Big data creates a new layer in the economy which is all about information, turning information, or data, into revenue. This will accelerate growth in the global economy and create jobs.” As the heart of Big Data, predictive analytics takes in disparate sources of data and extracts insights to provide accurate forecasts of future outcomes. PA is critical for providing a complete view of customer behaviors and trends. And businesses that learn to capture these insights will increase revenue, cut costs, and stay ahead of the competition.


So let’s pull together the main takeaways that should be included in your predictive analytics strategy going forward.




  • Identify your business problem: What is the primary question you wish to solve? For instance, you may decide: “We want to identify the top 10,000 customers who are most likely to respond to our mailers.”


  • Understand your data: Start by asking some basic questions: where does the data reside, who owns it, how far back, and in what format?


  • Clean your data: Disparate sets of data often contain duplicate, wrong and missing values, and other inconsistencies. Your data sets will need to be validated or “cleansed” to ensure optimal outcomes. Leverage available software tools to make this process less painstaking.


  • Build & evaluate your predictive model: This is a multi-faceted process that involves identifying the time frame of your study, choosing the dependent/independent variables, selecting the right methodology, and then building and testing the model.


  •  Deploy the predictive model: The journey has been a long one, but don’t stop short of rolling out the predictive outcomes to the rest of the organization. Deploying your model means sending live data to your predictive model and then feeding the results to your favorite data visualization tool. The results will be worth it!





Post Tagged with

About Jeffrey Walker

Jeff is a business development consultant who specializes in helping businesses grow through technology innovations and solutions. He holds multiple master’s degrees from institutions such as Andrews University and Columbia University, and leverages this background towards empowering people in today’s digital world. He currently works as a research specialist for a Fortune 100 firm in Boston. When not writing on the latest technology trends, Jeff runs a robotics startup called, along with oversight and leadership of - an emerging market assistance company that helps businesses grow through innovation.