In my last blog, we talked about leveraging predictive and prescriptive analytics to drive new value-added insights into the business. The focus now shifts to a hot topic in finance analytics today – machine learning (ML). What is it? Why is it important? What value does it bring? What is available today?
The first, and perhaps the most important step, is to define what machine learning is. According to SAS, an analytics partner of Teradata, “Machine learning is a method of data analysis that automates analytical model building. It is a branch of artificial intelligence based on the idea that systems can learn from data, identify patterns and make decisions with minimal human intervention.” Applying machine learning techniques to finance analytics is a growth analytic opportunity.
The SAS white paper “The Machine Learning Primer” states “Machines learn by studying data to detect patterns or by applying known rules to:
- Categorize or catalog like people or things
- Predict likely outcomes or actions based on identified patterns
- Identify hitherto unknown patterns and relationships
- Detect anomalous or unexpected behaviors”
Statistical models that predict results do not always fall under the umbrella of ML. Some models can analyze data and predict results based on patterns, trends, or other attributes, but do not “learn” from the past predictions or patterns or leverage machine learning techniques. However, these models are still are very useful in forecasting and planning of revenue and costs and are flexible in the amount of detail that is ingested.
ML works best for finance professionals in the areas where predictions of the future based on historical data are needed and when learning and refining the model by comparing the results to actuals drives improved performance. Although ML is based on the premise that updating the models can be done automatically, human intervention is often required. As the models mature, the less human intervention is required.
ML opportunities for finance professionals exist in many areas, the most obvious being planning and forecasting. The ability to provide insightful forecasts based on historical trends, internal attributes and external factors will improve accuracy and timeliness. Creating predictive models that can be updated to improve accuracy based on forecasted versus actual results transforms the planning process and makes it more accurate as time goes on.
Another potential area for machine learning is audit and compliance. The ability to update models as policies and regulations change lowers risk in the organization. For auditors, the ability to identify anomalies in transactional data enables them to identify potential issues as they happen and automatically. This includes outlier transactions, those that fail pre-established business rules or statistical patterns falling outside of ‘normal’ transaction thresholds.
One ML area outside of predicting the future is the ability to smartly perform cluster and or pattern analysis, which leverages past results to create smart clusters of data sets which have similar traits and produce predictable results. For example, in accounts receivables, we may want to leverage the historical payment patterns of customers to help identify which ones pay on time vs. those who are late. This enables business users to identify what pricing and contract terms should be used based on the ability of the customer to pay on time (or at least in a predictable manner).
Ultimately, the benefits from ML are realized in the business value provided to the customers and end users. The ability to predict revenue and costs based on data science and historical factors provides a better understanding of the factors that drive bottom line profitability. Audit automation enables the internal and external auditors to look at a wider sample of transactions and identify anomalies that may otherwise go undetected. Regulatory and compliance issues are easier to detect and can be proactively addressed before they become a bigger issue or result in penalties and fines. ML provides true business value and shifts the paradigm of the finance teams from reporter of history to provider of strategic analysis.]
Remember, CFO Analytics is made possible by following the critical steps of building the finance foundational model, automating manual processes, developing strategic analytics and leveraging the tools available to drive bottom line results that provide an ROI to the end users. By doing this, finance is transformed into an analytics factory that the business can rely on to understand profitable growth opportunities.
David Rosal is a Senior Industry Consultant for Teradata, focusing on helping customers through financial transformations including numerous Fortune 100 companies in Financial Services, Retail, Hospitality, Travel & Transportation and Manufacturing. He provides thought leadership on how to leverage integrated financial and non-financial data to drive innovative insights to improve performance and profitability through the use of data and analytics.
View all posts by David Rosal
David has more than 35 years of experience in the finance and management accounting space including technology, food service, retail and banking. He possesses a unique mix of finance, operational and technical skills across multiple industries and excels in the development of strategies and solutions that improve profitability and performance through the power of data. He has a BS in Accounting, MBA in Finance and is a registered CPA in the State of Illinois.