The Altis Buzz – December 2017
As we head off for a couple of weeks of recuperation, I would like to share a few of our most recent Insight items, including three Artificial Intelligence (AI) projects and the key blog posts over the last couple of months.
AI 1: Predicting product failures for a white goods manufacturer. Using our 3 week process, we identified a single product type as our test case and achieved over 85% accuracy of our predictive model. The next step is to scale the approach out across more products and adapt business processes to be prepared for better early intervention of wear out failures, which will increase customer satisfaction and decrease response costs. As a couple of reference points, the solution is hosted in the cloud for less than $200/month and for those statisticians, we tested Random Forest, Naïve Bayes and XGBoost, with XGBoost selected in the end.
AI2: Proof of concept to improve Call Centre Routing. 76% of 18 to 29 year olds prefer communicating via chat/text to their service providers. Our solution uses natural language processing and machine learning to classify complaint type, priority level, and customer tone to more effectively route and prioritise the incoming conversation. The objective is to improve customer satisfaction and time to close.
AI3: Lastly we are using machine learning, specifically Artitical Neural Network (deep learning) in Tensorflow to reduce the costs of manhole cleaning. Utilities spend millions a year in proactive and reactive maintenance of manhole covers to prevent or rectify blockages. Leveraging an IoT imagery system, we helped our client prove that we could train a solution to identify dirty manholes.
If you are interested in hearing more about our three week AI process to get your organisation started on its AI/ML journey, give us a call.
Key blogs from the past couple of months
Hope you enjoyed these insights, have a great break.
View a printable version of this newsletter here