Forestpin: Reducing Operational Risk by Identifying Outliers in Data

The massive influx of data has significantly given rise to outliers making data prone to threats and risks. It is becoming ever important for organizations to mitigate and manage operational risk by identifying outliers in data. These outliers are often the result of human errors, process issues or a result of deliberate manipulation by employees. Statistics from the Association of Certified Fraud Examiners (ACFE) show that companies having a fraud have lost 5 percent of revenue and it takes around 18 months to detect the fraud. Hence, it is essential to have visibility for correction of these outliers to improve quality of data and mitigate data-driven operational risk. Set out to solve the problem of risk mitigation, Forestpin created an analytics tool for customers enabling them to identify outliers using forensic data analysis. The company also provides a Risk Engine that alerts users on highest risk transactions on a daily basis. “The motto of the company: As our name suggests, similar to the phrase “needle in a haystack” we help our customers find a “pin in the forest” or outliers in the data,” says Ransith Fernando, co-founder and MD of Forestpin.

As a self-service forensic data analytics tool, Forestpin Analytics allows clients to very quickly analyze their data with simple visualizations of complex mathematical and statistical tests. The user can simply upload data or access a server connection, configure the default dashboard as required and the Forestpin’s tool provides the desired outcome. The company provides highly flexible dashboards to users where data can be easily fed through copy and paste or uploading data. In addition, the platform offers a field for verifying the data and generates a default dashboard to kick-start the analysis. The dashboard is completely configurable where a user can move analyses, decide on what he/she wants to see, make some analyses larger, add or remove analyses based on our predefined list of analyses and have multiple dashboards. Once the user is happy with the dashboard, the entire configuration can be saved. This allows for reusing the dashboard on a future date with more data or for sharing the dashboard with another user. For the enterprise customers, data automatically flows into the server, and a default dashboard is given, which once again is completely editable by the end user, and all configuration is local to the user and can be saved.

Once the data is fed, the data analytics tool provides many tests such as time series, correlation, relative time factor, first two digit tests, duplicates, composition, quadrant and stratification which highlight areas of risk or opportunity. In addition, clients can also benefit from BI tests such as the group, ageing, distribution, and box view. These provide for summarized information with data mining capability. “This is a proactive function of analyzing data to identify anomalies or garner business insight,” says Dilanke Hettiaratchi, co-founder and director of Forestpin. In addition, the company understands the challenges associated with analyzing high volume data from multiple data points and offers Forestpin Server that uses a file-based storage mechanism, 64-bit architecture, and in-memory computing. This helps with the response times and user experience. The company also provides the capability to combine data from multiple sources within Forestpin.

Forestpin’s Risk Engine works as a part of the enterprise system in the background where the data is extracted on a predefined basis, and a series of anomaly detection computations are run, resulting in a risk score on each and every transaction. The engine also computes a dynamic threshold based on the alerts sent in the last 30 days and the number of alerts a user can consume per day on average where the transactions above the threshold are alerted either in digest form or individual alerts. The engine’s dynamic workflow with authorization levels permits users to comment or change statuses based on their roles. The updated statuses are used by the learning mechanism to identify the risk factor to the business. If an alert is marked as not useful, then the Risk Engine will stop highlighting that type of risk over time. If an alert is marked as keep alerting, human error, process improvement or found manipulation then the learning mechanism increases priority for the type of alert. Interaction on the alerts is also captured for our enterprise customers on Forestpin analytics to ensure businesses are attending and investigating their alerts. “This is a reactive function to highlight risky transactions based on a predefined period resulting in financial savings by highlighting operational risk related to human error, process improvements, and manipulations,” says Hettiaratchi.

With Forestpin’s sophisticated tools, customers have changed about 0.7 percent of transaction values based on alerts, which is a significant value for an enterprise. The results can be calculated by using the alerted and changed statuses making the ROI very clear and measurable. With a complete toolset to cater to almost any scenario of data-driven operational risk, the company is working towards empowering customers to configure the tools as per their needs. As a future endeavor, Forestpin is also on focusing on opportunity gain/loss. “To get there, we are currently doing a lot of research on machine learning and applying different algorithms to identify the best mechanism to get there.” The company is also working with its existing customer base to identify the critical factors to identify opportunity within their businesses. “We do believe a tool or advising/consulting to a potential customer on factors which could either improve their turnover or profit will definitely be in high demand and we are constantly working on that,” concludes Fernando.