The Data Analytics Blog

Our news and views relating to Data Analytics, Big Data, Machine Learning, and the world of Credit.

All Posts

How To Get Started With Machine Learning

November 2, 2016 at 12:41 PM

The benefits have been recounted many times, but now that Machine Learning has the business world’s attention, how does one get started?  Moving into the machine learning space can be somewhat daunting, but we hope this blog post provides some guidance that you will find helpful. 

Machine Learning has been the topic of many of our blog posts, as well as articles in the media thanks to its ability to predict outcomes and automate decisions and thereby improve overall operational efficiency.  Just to clarify, Machine Learning is a system that has the ability to self-train the underlying predictive models by looking at recent data.

Still not 100% clear? Read our blog post What is Machine Learning?

Before getting started: Build upon a solid foundation

In this post, we are assuming that you have some predictive models in place and are using and tracking these in a couple of areas of your business.  Hopefully you or your team have developed these models internally and understand the concepts associated with predictive analytics, i.e. data scrubbing and data validation, building the models and validating them. 

The Brilliance of Ensemble Modelling

Before we describe our recommended approach to getting you to a fully-fledged Machine Learning system, let us describe a fundamental concept of predictive analytics, called Ensemble Modelling.  The basic concept is that one can combine various models that have been constructed using different algorithms into one stronger model.  

The brilliance of this concept is that one can use a stable, trusted model as the foundation and apply one or many real-time models on top of this - safe in the knowledge that the final model will not be worse than your ‘foundation model’, providing you use a tried and tested algorithm in constructing the combined model.  It is recommended that the models that get combined should be fundamentally different in order to capture different predictive patterns within your data.  A good example would be to combine a logistic regression model with a neural net model. 

This approach therefore gives you the best of both worlds: confidence that you will achieve at least what you had before (which has been proven to be reliable), but with the ability to bring in a model that captures more recent and different trends in the data. 

Getting Machine Learning into your business

With your foundation models in place, here are some pointers that will allow you to migrate your static predictive models to more dynamic self-training ones: 

  • Find an area of your business that you have identified as a good area for testing Machine Learning.  A good example would be in a call centre environment where there is a lot of movement and where changes occur on a regular basis – always an area that is ripe for Machine Learning
  • Package the required data and build a model as you normally would have done in the past (i.e. your foundation model).
  • Deploy this model and track it over time to ensure that it is performing as expected (you may already have such a model up and running)
  • Once you are satisfied that you have a model that you can trust, build one or more models by looking at more recent data, using different algorithms available in the tools that are at your disposal
  • Ensemble the new models with the foundation model.  This will dramatically reduce the risk of over-fitting the final model, whilst keeping it fresh by incorporating trends observed in recent data

Read our blog post Making the move from Predictive Modelling to Machine Learning.

How did it perform? 

We feel that the intention of a Machine Learning model is not to outperform the original static models, but rather to maintain consistent performance by bringing in recent data, instead of degrading over time, as is often observed in static models.  

Also note that, if the Machine Learning model does not add significant lift over the foundation model, it does not mean that Machine Learning approach should be shelved for good.  Various factors may be at play, such as how the dynamic models were constructed, the transient nature of the data, and the reliability of the new data. 

Assuming you are happy with the results (i.e. your Machine Learning models are outperforming your static models over time), you can now start considering incorporating a streamlined Machine Learning system which will contain the following elements: 

  • a platform from which all the required data will be processed,
  • a streamlined model building system, and
  • a streamlined and real-time tracking system to monitor the performance of all your Machine Learning models.

If you are ready to start the journey of bringing Machine Learning into your business or business area, learn more about our Machine Learning as A Service offering, Genius. It might make the path to getting started a lot shorter and cost-effective. Good luck and here’s to a long and fruitful Machine Learning journey.

 

predictive analytics guide

Image credit: Designed by Freepik

Robin Davies
Robin Davies
Robin Davies is the Head of Product Development at Principa. Robin’s team packages complex concepts into easy-to-use products that help our clients to lift their business in often unexpected ways.

Latest Posts

[Slideshare] How To Make Your Business Data Work For You

Common barriers to success: Skills shortage: data scientists are in high demand and in low supply. Companies lack the skills to develop advanced data analytics or machine learning applications. Cost: recruiting and building up or training a team, as well as infrastructure costs are immense. Inefficiency and low ROI on: acquisition campaigns; re-activation and retention campaigns; outbound sales calls and debt collection. Resulting in: No or ineffective use of data. High cost to get insights from data. Low returns from campaigns. What’s the alternative? Machine Learning as a Service (MLaaS): removes infrastructure skills and requirements for machine learning, allowing you to begin benefiting from machine learning quickly with little investment. Subscription based pricing, allowing you to benefit using machine learning while minimising your set-up costs and seeing returns sooner. Answers as a Service: Use historic data and machine learning to allow answers to increase in accuracy with time. MLaaS with predictive models pre-developed to answers specific questions: Genius Call Connect: What is the best time and number to call customers? Genius Customer Growth: Which customers are most likely to respond to cross-sell? Genius Re-activation: Which dormant customers are worth re-activating? Genius Customer Retention: Which customers are most likely to churn? Genius Leads: Which contacts are likely to respond to my campaign? Genius Risk Classifier: Which debtors are most likely to pay or roll? Benefits of Genius: Quick and cost-effective ability to leverage machine learning: Minimal set-up time Minimal involvement from IT Subscription based service Looking to make your data work for your business? Read more on Genius to see how it can help your business succeed. 

5 Must-Join Facebook Pages For Data Science, Machine Learning And Artificial Intelligence In 2019

While LinkedIn has traditionally been thought of as the business or work focussed social platform, Facebook has been making headway into gaining market share in the space as well. With company pages and groups, Facebook is catering to every interest and aspiration that people might have – and combining that with their social interactions and news sources. Facebook aims to give users a one-stop-shop experience, and it’s very good at doing it.

Our 2018 Customer Acquisition And Engagement Blog Roundup

Our final roundup this year covers two of our main topics: customer acquisition and customer engagement. We’ve not covered these topics in depth this year, and so decided to combine these two to provide a roundup of the best of both.