Understanding the Advantages and Disadvantages of Advanced Analytics

A law enforcement agency might use predictive analysis, looking at statistical data and algorithms, to determine crime rates, while a public health agency might use social network analysis to determine a disaster’s impact on different communities. Other forms of advanced analytics are geospatial analysis, text mining and network analysis.

Cybersecurity is a top federal priority, and agencies want to identify threats before breaches occur. To improve that capability inside networks, agencies must remove information sharing bottlenecks without compromising security.

Implemented properly, advanced analytics helps agencies identify organizational inefficiencies and leads to cost savings, but there are disadvantages as well.

MORE FROM FEDECH: How can agencies optimize for data ingestion at the edge?

The Disadvantages of Advanced Analytics

Advanced analytics comes with a high upfront cost, and agencies’ budgets aren’t unlimited.

Data quality issues need to be addressed before implementing advanced analytics, and there’s no standard set of technologies for collecting, offloading and protecting data.

Deploying and maintaining the proper infrastructure for advanced analytics requires paying highly skilled personnel, which is the biggest hurdle of them all. Technical experts need deep familiarity with their agencies’ data and the analytics available to them.

Government left industry to drive development of advanced analytics capabilities and is now skeptical about adopting the technologies. Many people consider them intrusive due to the prevalence of cameras in countries such as the United Kingdom and the portrayal of video analytics on TV and in movies.

The reality is that video analytics is already in widespread use in the U.S., not for tracking people’s movement but for analyzing traffic patterns to look for ways of alleviating congestion.

That’s not to say that agencies shouldn’t be mindful of data security and privacy concerns when handling large volumes of data. Agencies must choose the right algorithms for analyzing their data lest they get flawed data on the other side, and should consider the consequences of the data being misplaced, stolen or used improperly.

Compliance with government regulations is critical. When deploying advanced analytics, the Department of Defense must comply with Title 10, outlining the role of armed forces, and Title 50, guiding intelligence activities. Some agencies must also comply with state and local requirements.

DIVE DEEPER: How effective federal data sharing supports citizen services.

Early Advanced Analytics Adopters

All of that said, the Department of Homeland Security now uses advanced analytics to monitor social media for trends and threats internationally, the Centers for Disease Control and Prevention to flag disease outbreaks, and the General Services Administration to streamline procurements.

Agencies across the board are starting to adopt the machine learning and artificial intelligence (AI) capabilities necessary to perform advanced analytics.

Several agencies within the intelligence community were rapid adopters due to their need to analyze information quickly.

The National Oceanic and Atmospheric Administration also began using advanced analytics quite early to determine weather patterns, though its models remain far from perfect. While the agency can’t predict a hurricane’s entire path, it can identify probable movements. Information like that in the hands of emergency response personnel can help save lives.

EXAMINE: Why zero-trust architectures should include data protection and cyber recovery.

Starting with Advanced Analytics

Agencies beginning with advanced analytics must first plan for the outcome they want. That helps them identify the data they need, how to ingest it and what the constraints are for AI to use it effectively. Most CIOs have an idea of their desired outcome but lack a refined requirement to serve as a roadmap.

What advanced analytics solutions an agency settles on will depend on the amount of data they’re trying to process and from how many sources, but they’re similar. Data sets and learning models are what will vary from agency to agency.

Customizations to those models are where technical talent comes into play, as constant data set adjustments will be needed. If you think about it, Google is a large data set filter with search terms, but most of the time more research is needed to find exactly what users are looking for.

Tools like OpenAI, ChatGPT and GPT-4 have shown promise and are pretty predictive, but they’re still not where CIOs need them to be. That’s why Microsoft is trying to integrate Bing and Google into all of its platforms after seeing progress with Google Workspace.

For now, CIOs need to negotiate the gap between getting the right datasets and achieving desired outcomes.

This article is part of FedTech’s CapITal blog series. Please join the discussion on Twitter by using the #FedIT hashtag.