Augmented Analytics: Enabling Portfolio Managers in Smart & Effortless Decision Making
Co-authored by Akshay Arora
“The goal is to turn data into information, and information into insight”
-Carly Fiorina.
Extracting appropriate and adequate information from ample data is key for any business. Given the rate at which velocity and variety of data are multiplying, users find it difficult to consume the excessive information thrown to them through traditional BI systems.
The current business environment is dynamic and is changing at a rapid pace. Consumers behave differently based on various scenarios and thus, it is very challenging for a business user to keep track of an ever-changing environment. Just being aware of the change might not help users to take adequate actions unless provided with reasons for changes and their impact on the overall business. Hence, there is a dire need for a system that could alert users about the dynamic business environment by monitoring KPIs and generating automatic alerts with proper root cause and impact.
BRIDGEi2i’s AI-powered Watchtowerᵀᴹ, an augmented analytics solution, leverages advanced machine learning algorithms to identify changes in the business by monitoring KPIs across different dimensions and alerts users with appropriate root cause and impact. It helps business leaders to focus on things that matter most i.e. taking adequate actions rather than spending time figuring out what went wrong.
Here’s a case study that illustrates the working of the Watchtowerᵀᴹ!
A little info about our client: We worked with a US-based Fortune 100 company that’s majorly involved in investment banking and financial services in effective portfolio tracking.
Understanding the business context: Portfolio managers often track over 100s of KPIs related to portfolio performance and macro-economic factors across various dimensions to reduce the risk of investment in various portfolios. Any wrong investment could have a major impact on the company’s return on investment.
The BRIDGEi2i Solution
Our Watchtowerᵀᴹ helped identify geographies where the risk of investment is high by monitoring over 150 KPIs and macro variables in real-time. This was done by identifying anomalous patterns across various metrics and identifying the appropriate root cause. Potential business impacts of change in macro variables and portfolios were also highlighted and plotted.
Use of On-us and Off-us Data
This solution uses both On-us data as well as macro-economic data. On-us data includes loan performance metrics like default 30+, default 60+, bankruptcy, Total outstanding, delinquency rates, etc. From Off-us data macro-economic variables like unemployment, growth, inflation, etc and bureau variables were used.
Use of ML-based Algorithms for Detecting Anomaly/Change in Pattern
The Watchtowerᵀᴹ uses proprietary ML algorithms to learn temporal & spatial patterns from metrics dynamically and identifies any abnormal/anomalous pattern arising in data. A suite of algorithms were applied on the data and suitable algorithms were identified for each KPI based on the distribution of data. The study of comparative performances of geographies can help portfolio managers to balance portfolios across geographies.
Finding the relationships between the KPIs
The causal model helped in identifying relationships between KPIs across different levels of hierarchies. Advanced statistical algorithms were used to identify the strength of relationships. It helped to determine the relationship between macro variables and default rates in various portfolios like retail, auto, house etc. Causal relations and anomaly detection enabled root cause drill down and predictive alerts generation.
Finding Root Cause and Impact of Anomalies
Causal relations, business hierarchies and output of pattern/anomaly detection modules are stored in graph database. Proprietary graph storage and traversal algorithms were used to cater custom needs of root cause exploration and impact estimation. Graph database helped query large sets of relations and business hierarchies faster and at scale. Root causes were ranked based on their impact.
Each alert could have an impact on upstream KPIs or hierarchies. It was important to quantify the impact of change on the overall default rate. System uses statistical models to quantify the impact on change in KPIs on overall default rate.
Business Impact
Watchtowerᵀᴹ helped portfolio managers track 150+ KPIs across 30 different regions, loan products, and 10 other dimensions effectively. Watchtowerᵀᴹ solution helped stitch parts of analysis which assisted managers in taking action without spending much time looking through dashboards enabling portfolio managers to take faster and accurate decisions.
Watchtowerᵀᴹ used chart recommendation algorithms to suggest the right charts to users for exploring alert & root cause better. Alerts in the form of text helped usage of analytics by broader sets of users and yet making uniform interpretations of alerts.
BRIDGEi2i’s Watchtower can be applied to any functional and business areas with customization. To know more, you can read about how we helped a global CPG company with the Pricing Watchtowerᵀᴹ.
You can also read Why Traditional BI System are Not Powerful Enough for Business Leaders?
I would like to thank my co-author Akshay Arora for his contribution in writing this blog.
