A finance leader in the greater Chicago area, Erich K Squire has served as an independent business consultant since stepping down as the senior financial officer of Century Aluminum in 2017. As part of his efforts to provide comprehensive and innovative strategic and analytic services, Erich K Squire has developed a keen professional interest in machine learning.
%20(1).jpg "Erich K Squire Examines")
Machine Learning in Financial Modeling
Machine learning is a
subcategory of artificial intelligence (AI) that has grown by leaps and bounds
over the past couple of decades thanks to exponential advancements in both
computer software and hardware. While AI involves developing technology that
allows machines to convincingly mimic human thought and behavior, machine
learning involves developing technology that can automatically learn from past
sensory data and informational input without the direct influence of human
programmers.
Leveraging the power of highly sophisticated
algorithms, experts in the fields of business and investment have employed
machine learning to create progressively detailed financial models. These
models have proven effective methods of predicting the success of a wide range
of company development strategies and asset management options.
Interpretability Understood
To bridge the gap between the conceptual
methodology behind the machine learning algorithm and both its implementation
and value in the real world, AI scientists have created a concept called
“interpretability.” Briefly defined by Interpretable
Machine Learning author Christoph Molnar, interpretability can be
identified as the degree to which a human being can understand the cause of a
machine learning decision or as the degree to which a human can dependably
predict the result of a machine learning model.
The Supreme Importance
of Interpretability
Even those who place complete faith in the
accuracy of a specific financial model must recognize just how important
interpretability can be when it comes to establishing and reinforcing trust in
the machine learning processes that developed that financial model. Because
human beings are often hesitant to rely on machine learning models when it
comes to making potentially costly decisions, business leaders and investors
may need extra information or reassurance from those who develop and utilize
these models.
Just as important as the issue of trust,
contestability keeps financial modeling machine learning in check by making it
possible for knowledgeable people to appeal the decisions that those models
favor. Black-box proprietary recidivism predictors like COMPAS have garnered a
significant amount of criticism due to the lack of contestability integrated
into their processes.
Straightforward and contentious approaches to
interpretability that emphasize transparency as well as verification and
substantiation can help alleviate many fears involving trust and
contestability. However, safety concerns may persist even if relatively minor
shifts occur between the model in conception and the model in deployment.
Analysts who use machine learning can further the adoption of this technology
by explaining the representations of their financial models and/or highlighting
their most relevant features.
For More Information
If you want to learn more about machine
learning and its place in modern AI and financial modeling, contact Erich K Squire in his Chicagoland office
today. He has an extensive background in financial analysis, forecasting, and
modeling with a focus on the latest technological tools.
read more: Lanyard Style and Size Guide