In the the final article in our current series of articles from field service solution provider Astea focussing Artificial Intelligence and Machine Learning,  we see how applying the Machine Learning can lead to better service resolutions... 

Is Machine Learning a key topic for your organisation?!

There is a detailed white paper on this topic authored by Emily Hackman and Liron Marcus which is available to fieldservicenews.com subscribers within our premium content library... 

Sponsored by: 

Data usage note: By accessing this content you consent to the contact details submitted when you registered as a subscriber to fieldservicenews.com to be shared with the listed sponsor of this premium content who may contact you for legitimate business reasons to discuss the content of this content.   

Congratulations, if you have been following this series of articles across the last few weeks you are now a cognitive technology guru! (If you haven't don't fret! we've written a whole white paper on the topic which is available to fieldservicenews.com subscribers.)

In the series we've tried to give you an understanding of how and why the field service industry is applying these new technologies, and you see the writing on the wall: positive results for your customers, your employees and your financial success.

Now it’s time for you to go through the machine learning process yourself in order to answer the most important question ever asked in a field service scenario: “What is the best resolution to a given problem?”

If we break this question down into smaller operational components, we first need to maintain a set of wide range problems represented in groups which share similar characteristics. We then label these groups. Let's say, in our case, how the problems were fixed or repaired. Then, we need to create a learning process that labels a new problem based on its characteristics.

The Learning Process

The learning process requires “real world” raw data and actual business experts. In our example, this data will be historical incidents and service requests that could most likely be predictive. The business experts are the people with extensive knowledge of hands-on technical support and service (e.g. field service managers with years of service and product experience.)

The next step is called data preparation. This step is required to screen the raw data, remove duplicate or irrelevant records, handle missing data attributes, and extract indicators and features that are needed in the learning process.

"Free text", such as the problem description of these examples, is then extracted into a bag-of-words model. This is a representation of the description text with meaningful information (disregarding semantic relationships in the sentences).

The next step is called the training process. This is where the actual learning happens. Different types of learning algorithms are applied to a data training set (a subset of the entire data to learn from).

The learning process requires “real world” raw data and business experts. In our example, this data will be historical incidents and service requests that could most likely be predictive.

The choice of those metrics influences how the performance of the machine learning algorithms is measured and compared. Accuracy is one of the metrics to evaluate and classify models. It calculates the number of correct predictions made by our training model over all kinds of predictions being made. Another metric, called the confusion or error matrix, is a two dimensional table that visualizes “actual” versus “predicted” results for different questions, called the “classes”. It shows if a model is confusing two classes (e.g. commonly mislabeling one as another).

Aptly named, the confusion matrix visualizes “actual” versus “predicted” results for different questions, called the "classes". It shows if a model is confusing two classes.

Do you want to know more?!

There is a detailed white paper on this topic authored by Emily Hackman and Liron Marcus which is available to fieldservicenews.com subscribers within our premium content library... 

Sponsored by: 

Data usage note: By accessing this content you consent to the contact details submitted when you registered as a subscriber to fieldservicenews.com to be shared with the listed sponsor of this premium content who may contact you for legitimate business reasons to discuss the content of this content.