Founded in 2012, Colombo-based data analytics firm combines mathematical programming and data visualisation skills to develop easy-to-use analytics applications. Forestpin allows users to analyse enterprise transactions such as payments, invoices and sales orders to find outliers. The company takes advantage of new hardware technologies such as 64-bit computing and cheap memory, to run all analytics keeping the data in memory.
Analytics India Magazine interacted with Ransith Fernando, co-founder and managing director of Forestpin, who talked about how the company, is helping businesses in identifying anomalies using data analytics.
Analytics India Magazine: Tell us about Forestpin and various products that the company offers.
Ransith Fernando: Forestpin is specialised in software that detects anomalies. We have two products – Forestpin Analytics and the Forestpin Risk Engine. Forestpin Analytics is sold as single-user licenses or server-based licenses. And the Risk engine is sold as server-based licenses.
Forestpin Analytics single-user licenses are targeted at financial forensic data analysts, auditors and fraud investigators. We give them the ability to use specialised analyses without any scripting or dashboard creation. By using Forestpin Analytics they can upload a dataset and automatically get a fully functional editable dashboard. The server-based versions and the risk engine is targeted for companies interested in operational risk reduction.
AIM: How is Forestpin utilising analytics to provide solutions to businesses?
RF: Our analytics help companies find an unusual transaction that can be either due to a data entry error, process issue or manipulation of data. We help companies identify individual transactions that need correction. When it’s down to a single transaction that looks odd you can directly act on it.
AIM: What are the data visualisation services offered by Forestpin? How does it compete with other business intelligence and visualisation companies across the globe?
RF: The key differentiator is that Forestpin visualisations are not the end output. These are intermediate steps to drill down into the single transactions of interest. To find unusual transactions we have many analyses and visualisations. Some of the analyses and visualisations have been developed from scratch. We believe these have to be unique to our product. Our customers already had best of breed visualisation tools before they purchased Forestpin. These were complemented by the use of Forestpin Analytics and Risk Engine.
AIM: Tell us about Forestpin’s play in forensic data analytics.
RF: We have some well-known Forensic Data Analyses like the FTD (First Two Digit, which is based on Benford’s Law and the RSF (Relative Size Factor). We have implemented these with a twist in functionality. Some analyses are developed from scratch like our composition analyses, compare timelines correlation and quadrant timelines. These analyses are unique to our product.
In addition, we have analyses based on time series and correlation. Forestpin Analytics lets end users the ability to use these analyses without much technical knowledge and setup steps. To use a time series, a user just has to drill down from the red lines that are highlighted based on differences between the actual v/s the prediction. The user does not have to pick an algorithm or set any parameters or thresholds. Based on the uploaded user’s data we automatically give a usable dashboard. The objective is to enable a user who has no context to the data still find unusual transactions.
AIM: What are the challenges that Forestpin faces as a company?
RF: Managing cash flow is a challenge. We are bootstrapped so we cannot put a lot of money for marketing, which makes growth slow and the sales cycles for enterprise sales are long, which puts pressure on our cash flow.
AIM: Who are some of your clients? How have they benefited from the tools and services offered by the company?
RF: John Keells Holdings, the largest listed company by market capitalization in the Colombo stock exchange, uses Forestpin Analytic across 43 companies in insurance, supermarkets, hotels, travel, manufacturing and food industries. The largest privately-held company in Sri Lanka, MAS Holdings, also uses Forestpin Analytics in their division MAS Intimates.
We also have specialised Forensic Data Investigators using the product and it’s been adopted by Universities as part of their data analytics program. The licenses are given free to Universities and they love some of the unique analyses of Forestpin. Last year, our partners in Malaysia had a competition on Forensic data analytics in the Asia Pacific University.
AIM: Does Forestpin work with any Indian company?
RF: Not as yet. We are building brand awareness in India. We took part in the Inventicon Annual Corporate Fraud, Forensics India Summit in 2017 and in the Operational Risk Management Summit in 2018. We got a very warm welcome in India and it felt like home. We are building contacts in Mumbai, and currently looking for a sales partner who can provide local support.
AIM: What is the roadmap for analytics at Forestpin?
RF: We plan to venture into industry-specific solutions to selected user cases keeping the same code base. We have been concentrating a lot on technology, but what people want to know is what exactly we can do for them. Our roadmap is to have industry focused user cases in the insurance and manufacturing industries. We also plan to offer consulting services where we do an ad hoc analyses of a companies data.
AIM: What is the size and hierarchical alignment (both depth and breadth), of your analytics group?
RF: Our staff size is small in number and we have a flat structure. We have outsourced accounts and sales are through partners. HR filings of EPF, ETF and taxes are done by an outsourced company. The core products are done by our internal team and for the side projects and some specific tasks, we have hired contractors and interns.
We are strong on machine learning and have customers who have been benefiting from the machine learning components of our product on a daily basis for over 18 months. At present, we have four people doing research on deep learning.
We also support open source projects on machine learning. Currently, we are supporting a pet project of one of our co-founders on building a framework that explains what your code in the Jupyter notebook does.
Try deep learning using MATLAB