In this four part series, Jarlath Quinn demonstrates Smart Vision’s new Repeatable Application Template, Predictive Operational Analytics, showing how it combines market-leading predictive analytics software, ready-to-use application templates and a fully supported professional services package to enable rapid and effective implementation and deployment.
Using IBM’s flagship data science tool, SPSS Modeler, he works through a telecoms maintenance case study to show how sophisticated text mining can be applied to error messages and engineer logs, how a predictive model can be built to determine whether a maintenance task will require a part replacement, how to estimate which parts are likely to be needed or whether a simple reset will suffice, and how to identify asset sites most likely to generate a further error or task within 20 days of job completion.
Predictive operational analytics part one
In part one of this video series Jarlath Quinn explains how operational analytics works, what data sources may be utilised and introduces the example case study that forms the basis of the subsequent videos in the series.
Predictive operational analytics part two – SPSS Modeler demo
In part two of our predictive operational analytics video series, Jarlath Quinn introduces the IBM SPSS Modeler Software before showing how to explore the example data from the case study and carry out text analytics of engineer logs to categorise the information and create key fields for further analysis.
Predictive operational analytics part 3 – predictive part replacement
In part 3 of this predictive operational analytics guide Jarlath uses the results from the previous analysis stage to show how to build and assess a predictive model that identifies whether or not a part replacement will be required during a maintenance visit. We also see how to apply the model to new data so that predictions can be generated before the engineer is dispatched.
Predictive operational analytics part 4 – predicting task outcomes
In this final part of our predictive operational analytics video series Jarlath shows how we can use the predictive technology to go further by predicting the actual outcome of the visit including which part is likely to require replacement. In the final example we see how to build a model that predicts whether or not a site will require another unscheduled visit within 20 days.