How AI and Machine Learning Finds and Fixes Issues Faster

Applications are only as good as the trouble-ticketing systems that support them. What separates good application experiences from great ones lies in this question — How fast can I get issues addressed?

It all starts with a ticketing system and how issues can get in the queue and remediated as fast as possible. It’s an integral step in the application development process, but one many companies don’t consider upfront.

That’s not the case with the experts and developers at NTT DATA. It’s always top-of-mind as they architect and deploy client applications. For that reason, we created NTT DATA Advance Ticket Analytics, which relies on AI and machine learning to better categorize tickets and allow customers to proactively remediate issues before they hinder users’ productivity and elevate their frustration.

Companies need to know how many team members to deploy beforehand and uncover optimization opportunities to provide Year-Over-Year (YoY) productivity gains. In short, they need to go from the reactive to the proactive. Often, companies turn to industry-related research to plan for the future; this may work, but there’s a good chance it won’t. There’s too much at stake to hope it will address your specific needs. A solid ticketing system, such as NTT DATA’s Advanced Ticket Analytics, should provide key insights to allow companies to make changes with more confidence and accuracy.

Current ticket analysis — antiquated, slow, manual and error-prone

Traditional ticketing systems are rife with issues that prevent analysis from being done comprehensively and quickly. Whether opening a ticket to report trouble or request service, access or a change, current systems make analyzing them laborious and time-consuming. The need to cluster and categorize tickets often requires days to properly analyze and complete.

When a company needs to understand how and where they need to enhance the user experience, waiting days is anything but efficient. Here are some other issues related to the current state of ticket analysis:

• Categorization of tickets relies on token-based keywords and don’t take into account semantic meanings of content supplied by users.
• Text descriptions are lengthy, which requires time and effort to get to the heart of the issue.
• As ticket volume increases, so does resource utilization, which prevents personnel from focusing on other pressing initiatives.
• Previous engagements aren’t easily leveraged to proactively address future issues.
• Forecasting, ticket similarity and time series aren’t taken into consideration.

NTT DATA’s Advance Ticket Analytics — fast, accurate, intelligent

NTT DATA’s Advance Ticket Analytics uses AI and machine learning to supplant traditional, inefficient ticket analysis. Most ticketing analysis often relies on Excel/VB macro programs to analyze ticket data, which means classifying them is based solely on descriptions. This approach, however, only clusters about 15 - 20% of trouble tickets. This means the other 80% need to be categorized manually.

While this may not be daunting to a small company that gets less than 100 tickets a month, it becomes impossible for large enterprises, which can have tens of thousands of tickets. It leaves analysts having to make assumptions and hope they’re correct. It’s like a geologist unearthing a couple of bone fragments and assuming they’ve discovered a dinosaur. The assumption might be correct, but there’s an equal chance they won’t.

With AI and machine learning incorporated into trouble ticket analytics, there’s no need for assumptions. NTT DATA clients enjoy improved analytics through the collection of data that can be easily analyzed to deliver key insights fast. Recurring issues are brought to light more comprehensively, and sooner, which results in faster remediation.

With NTT DATA’s Advance Ticket Analytics, up to 300% more trouble tickets can not only be categorized, but more accurately. Analysis can be completed in a matter of hours, as opposed to days. Issues that need remediation and focused attention are brought to the forefront faster.

AI- and machine learning-driven ticket analytics provide insights that help users make decisions on automation and shift left levers. The insights are context sensitive and consider the specific towers (applications, infrastructure, et al.) and domains, such as healthcare, financial services, and others. They are designed to be trained to scale to specific towers and domains as more historical data is made available.

NTT DATA’s Advance Ticket Analytics is built and trained on the Dataiku platform, which uses up to 1 million training data sets to base categorization across towers and domains. The analytical models used are a mixture of clustering, forecasting, correlation, text analytics and time series. If NLP-BERT- is used, similar search models are leveraged. Power BI is used for visualization and presentation. And it can be hosted with Microsoft Azure, or any cloud platform, and can be scaled on demand.

To teach the machine learning element, over ten accounts of data are used to teach the system so it can learn and apply its findings to address other tickets This data is streamlined, cleansed, standardized and normalized based on different industries. Over 5,000 tokens from an existing rule-based engine are leveraged for the training, which delivers an accuracy level around 97% for shift left and automation levers.

NTT DATA’s Advance Ticket Analytics was built with three key components in mind:

1. Analytical/NLP Model-Based
By leveraging Natural Language Processing (NLP), data can be accurately deciphered, allowing for more comprehensive and accurate ticket categorization. Useful data can be uncovered that is far harder to analyze if done manually. It can even reveal opinions of users, something that can be easily overlooked if left up to an individual who is reviewing thousands of tickets manually.

2. Insight-driven output
Outputs provide customers with better, more targeted insights, giving them the ability to make better decisions related to resource and skill set allocation.

3. Learning based on actual, historical data
By leveraging machine learning, ticket categorization relies on historical data analyzation, becoming better and better with time.

Here are a few of the benefits clients are enjoying with NTT DATA’s Advance Ticket Analytics

• Forecasting of future ticket volumes
  NTT DATA’s Advance Ticket Analytics takes remediation from the reactive to the proactive, enabling users to prepare for upcoming events and staff up to ensure they’re at the ready. For instance, a batch system failure will likely create a precipitous rise in tickets, so finding out about it beforehand means remediation can occur prior to it affecting productivity.
• Root cause identification
  Customers quickly learn why an event took place, saving considerable time and allowing them to create and deploy steps to prevent them from happening again.
• Reduction in costs
  Customers can save both hard and soft costs, whether it’s in requiring fewer resources or better allocating them to work on other initiatives.
• Increased efficiency
  With AI and machine learning, client IT teams can glean key insights through predictive analytics to streamline incident management, improve workflows and better incorporate demand planning.
• Faster remediation
  Issues and trends are brought to light faster, enabling developers and solutions architects to address them before they result in frustrated employees and disgruntled customers. It’s a shift from reactive to proactive.
• Enhanced knowledge sharing
  While the categorization of tickets becomes smarter with machine learning, so do personnel by staying ahead of issues before they become disastrous.

Learn more about how NTT DATA Advance Ticket Analytics can help you proactively remediate issues before they hinder users’ productivity, reach out to the experts here. Then just click on the Contact an Expert button at the bottom. They’d love to hear from you.