Predictive Analytics in a Data-Driven World

Access to large bodies of unstructured and semi-structured information is growing, but the real question is: How do lawyers leverage the availability of this information in a useful way?

'Law firms don't think of themselves as data-driven,' says Professor Katz. 'They don't consider saving data across all their sources of information. But that's what modern management is about; they start by collecting data ' massive amounts of information ' and use it to generate sophisticated predictive models as a basis for making decisions.'

The most disruptive of all possible displacing technologies ' quantitative legal prediction ' can enable this process in the practice of law, and is likely to drive a substantial amount of the future innovation in the legal services industry.

How Does Predictive Analytics Work?

Predictive analytics is a blend of tools and techniques that enable organizations to identify patterns in data that can be used to make predictions of future outcomes. In business, predictive analytics typically take the form of predictive models that are used to drive better decision making. They find and measure patterns to identify risks and opportunities using transactional, demographic, Web-based, historical, text, sensor, economic and unstructured data. These powerful models are able to consider multiple factors and predict outcomes with a high level of accuracy.

The three functions of predictive analytics are:

1. Pattern detection;
2. Differentiation; and
3. Tying together relevant data from different data sets to form a conclusion.

In order to make a prediction regarding any number of unknown outcomes, a lawyer must examine data with an eye toward detecting patterns of behaviors and outcomes. This data arises from a variety of data sets that are often impossible to integrate for the purpose of comparative relevancy (e.g., social media data, court information, legal precedent, etc.), and present a variety of data types: structured, semi-structured and unstructured. By repetitively examining this data, lawyers, researchers and IT developers begin to learn and apply that knowledge across different classes of problems and reason by analogy. The exercise then is to create algorithms and incorporate data mining techniques to train computational functions to do the same thing, but with more ease and in less time.

e-Discovery is the first component of the legal process to experience a great deal of success in creating reliable outcomes by using keywords and coding methods that form relational bases. This enables pattern detection and differentiation among various types of data.

However, 'it is not correct to believe that it can't apply to other classes of problems, like what's going to happen in a case,' Professor Katz says. 'What are the features of a personal injury case that will drive the outcome? This insight can be obtained through data. Reasoning has always been the centerpiece of how legal judgments are made. The development of algorithms and use of data-mining techniques are making incursions into the industry, but you must look outside the industry to see how other it is being used in other spaces.'

Conclusion

In the end, predictive modeling will be limited only to the extent it relies on the limits of human creativity and analysis in enabling it to mimic the behavior of 'expert reasoners': What does it mean to 'think like a lawyer?'

While humans are amazing in their abilities to detect patterns, aggregation is a problem. How much data can a person consider? Machines have no such restrictions. They are not limited by the failings of memory loss, subjective perspectives, or narrow vision. They can proceed to perform high-level pattern detection, high dimensional similarity matching and analogical reasoning to produce predictions that will be reliable, at a fraction of the cost and time.

Donna Seyle is an attorney, writer and founder of Law Practice Strategy, an information center on the future of law practice and legal technology. She is also a member of the ABA-LPM's eLawyering Task Force Committee. Seyle may be reached at [email protected].

The use of predictive coding in e-discovery answers the need to automate document review for discovery purposes. Why is automation of this process now so necessary? Because Big Data is upon us, and the legal profession is as much affected by this mountain of information as is business. We need automation to make our way through this.

Prediction coding is one part of the process of quantitative predictive analytics, and the use of predictive analytics in the practice of law is charging well past its role in e-discovery. Think about it: how often is a lawyer required to make a prediction: Do I have a case? What is our likely exposure? How should this legal matter be priced? Who in the firm is best equipped to handle it? What's up with the judge? What's the best jury composition? These are the questions that can, and will, rely on human-trained technology designed to wrest these decisions from the frailties of human thinking.

Daniel Katz, assistant professor of law at Michigan State University College of Law, and co-founder of Reinvent Law Laboratory, explains that the circumstances creating this environment include the monumental increase of Big Data, the vast decrease in data storage costs, and advancements in computing power and learning. ReinventLaw identifies four pillars of innovation for the legal services industry: Law+Tech+Design+Delivery. As counsel and legal consumers act to drive down costs ' demanding value propositions and reduction of legal spend, the use of predictive models in legal analysis has fertile ground on which to develop and grow.

Access to large bodies of unstructured and semi-structured information is growing, but the real question is: How do lawyers leverage the availability of this information in a useful way?

'Law firms don't think of themselves as data-driven,' says Professor Katz. 'They don't consider saving data across all their sources of information. But that's what modern management is about; they start by collecting data ' massive amounts of information ' and use it to generate sophisticated predictive models as a basis for making decisions.'

The most disruptive of all possible displacing technologies ' quantitative legal prediction ' can enable this process in the practice of law, and is likely to drive a substantial amount of the future innovation in the legal services industry.

How Does Predictive Analytics Work?

Predictive analytics is a blend of tools and techniques that enable organizations to identify patterns in data that can be used to make predictions of future outcomes. In business, predictive analytics typically take the form of predictive models that are used to drive better decision making. They find and measure patterns to identify risks and opportunities using transactional, demographic, Web-based, historical, text, sensor, economic and unstructured data. These powerful models are able to consider multiple factors and predict outcomes with a high level of accuracy.

The three functions of predictive analytics are:

1. Pattern detection;
2. Differentiation; and
3. Tying together relevant data from different data sets to form a conclusion.

In order to make a prediction regarding any number of unknown outcomes, a lawyer must examine data with an eye toward detecting patterns of behaviors and outcomes. This data arises from a variety of data sets that are often impossible to integrate for the purpose of comparative relevancy (e.g., social media data, court information, legal precedent, etc.), and present a variety of data types: structured, semi-structured and unstructured. By repetitively examining this data, lawyers, researchers and IT developers begin to learn and apply that knowledge across different classes of problems and reason by analogy. The exercise then is to create algorithms and incorporate data mining techniques to train computational functions to do the same thing, but with more ease and in less time.

e-Discovery is the first component of the legal process to experience a great deal of success in creating reliable outcomes by using keywords and coding methods that form relational bases. This enables pattern detection and differentiation among various types of data.

However, 'it is not correct to believe that it can't apply to other classes of problems, like what's going to happen in a case,' Professor Katz says. 'What are the features of a personal injury case that will drive the outcome? This insight can be obtained through data. Reasoning has always been the centerpiece of how legal judgments are made. The development of algorithms and use of data-mining techniques are making incursions into the industry, but you must look outside the industry to see how other it is being used in other spaces.'

Conclusion

In the end, predictive modeling will be limited only to the extent it relies on the limits of human creativity and analysis in enabling it to mimic the behavior of 'expert reasoners': What does it mean to 'think like a lawyer?'

While humans are amazing in their abilities to detect patterns, aggregation is a problem. How much data can a person consider? Machines have no such restrictions. They are not limited by the failings of memory loss, subjective perspectives, or narrow vision. They can proceed to perform high-level pattern detection, high dimensional similarity matching and analogical reasoning to produce predictions that will be reliable, at a fraction of the cost and time.

Donna Seyle is an attorney, writer and founder of Law Practice Strategy, an information center on the future of law practice and legal technology. She is also a member of the ABA-LPM's eLawyering Task Force Committee. Seyle may be reached at [email protected].