Within the financial services sector there is a move towards digital convergence. This drove our client – a multinational financial services corporation - to launch an exercise to ensure readiness of their core network to support this convergence and also ward off threats from non-traditional players within the market. Their network estate was complex and over time has developed characteristics hindering their ability to meet future business needs. They sought our help to tackle this challenge.
We set about curating knowledge that lay within the current core network landscape. The environment was a conglomeration of systems “thinking in silos” and we invested significantly to create a single unified true picture of the enterprise IT. We then used this visualization to map out how various business initiatives were being served by this system. We also analyzed the competitor landscape to determine market trends that would influence future strategy and identify gaps within the client’s environment.
Based on this knowledge map, a future-state was drawn up to modernize the network technology landscape and a set of initiatives were defined and prioritized to arrive at the roadmap for change.
The key themes for the transformational blueprint were:
A large specialty apparel retailer with an IT landscape housing more than 400 applications, was looking to improve the availability of their IT systems. The IT environment was complex with applications built over decades using multiple technologies including mainframes, custom built thick client applications, web applications, and packaged applications. With interdependencies that were poorly documented, manual analysis of IT issues was taking time and great effort. There was a crying need for automation and faster resolution of issues faced by business partners using these IT systems.
Infosys leveraged Infosys Automation Platform (IAP) to analyze the historical data of the incidents, service requests, alerts and other IT support tickets. IAP used machine learning algorithms, natural language processing and text analytics to find recurring patterns and identify opportunities for automation.
Infosys then leveraged our support engineers’ deep expertise in the retail domain and in systems for warehouse management, e-commerce and inventory management to create a digitized knowledge base of issue resolution and service request fulfilment processes in IAP. The team then created intelligent software robots using IAP that leveraged this knowledge base to resolve IT service management tickets. These robots picked up various tickets, diagnosed issues, determined the resolution to be applied, and executed necessary actions.
Unexpected breakdown of equipment (even non-critical) in pharmaceutical manufacturing has severe, multi-dimensional impact. In addition to the costs of down-time and maintenance, there are also potential implications on manufacturing process quality and even drug safety. Given this, a multinational pharmaceutical company was keen to leverage big data technologies and analytic methods to develop a predictive maintenance approach for a broad range of equipment in their plants.
Key goals were to:
To develop, test and arrive at the optimal predictive analytics approach, Infosys focused its knowledge curation efforts on a specific set of equipment – a set of reactors and upstream de-gasifier (cylinders) from one of the pharmaceutical’s plants.
Data extracts for 18 months were analysed to identify major breakdown events to be predicted (the independent variable). Programmable Logic Controller (PLC) system data stored in the database from sensors for pressure, temperature and weights were used as another set of predictors or independent variables. In addition, PLC alarm patterns (such as the count of alarms/hour) were also used as predictors.
Leveraging Infosys Information Platform, a logistic regression (binomial logit) model was trained using a portion of data, retaining the rest of the data to validate and test the trained model. The model was developed for predictions ahead by 1-day and 2-days. Model score cut-off for predicting a potential breakdown was chosen to balance the capture rate vs false alarm percentage as these two represent a trade-off.