Improving the Quality of End-Of-Day Reporting With Data Science Tools

IMPROVING THE QUALITY OF END-OF-DAY REPORTING WITH DATA SCIENCE TOOLS

  • SuriBabu Boddu, Golokhvastova Daria
  • Published: 13 August 2021


Overview

End-of-day (EOD) reporting is the most important part of any banking application maintenance with respect to BAU operations. Banks use automated EOD (or intraday) job plans to generate reports for regulatory monitoring, business users, downstream stream systems etc., and to deliver them to stakeholders for various purposes. Such jobs or reports can be classified into:

  • Reports delivered to downstream applications for further processing and usage in downstream application
  • Reports delivered to regulators as part of mandated regulatory compliance
  • Reports delivered to management for further checks and decision making.

‘Report’ as a term has been used for decades to represent data file delivery. This definition in the functional sense, however, simply means published data, as opposed to processed information which can facilitate the decision-making process.

The problem: EOD reports present partially processed data, not usable for decision making

At present, the EOD reports are usually delivered to different stakeholders in formats XLS, CSV, DAT, or TXT as partially processed data.

Fig 1: Sample report format in current setups

A sample of a report in the figure above represents data as requested by the end user. In order to offer the end-user information which is useful in decision-making, however, further processing steps are required for making the process time-inefficient and including the possibility of human error.

In the reminder of this blog, we will look at the possible solutions to this optimization problem. It is important to understand what the end-user use the reported data for and the purpose of the reporting, i.e. what is the end-result the user wants to achieve with this process step?

Suggested solution: Data science methodologies

An institution and its stakeholders can achieve high benefits from integrating data science methodologies into the EOD reporting. Data science methodologies can be used to clean, analyse and visualize the data for better representation of the final summary, providing useful information to the user.

Complete business logic for analysing reports and suggesting feasible solutions automatically, which the end-user would otherwise need to apply manually, can be implemented using Python (or any other programming language) and Data Science methodologies.

With this solution, we can generate an enhanced report, which will include new sections with insightful information, in addition to the data that the end-user has been receiving so far. The suggested report generation would look as below:

       

Fig 2: Reporting with Data Science visualization

The data in the report has been analysed, processed, and summarized in a way that is easy to understand. This representation would enable the end-user to make faster and better-informed decisions based on the delivered facts and figures.

Implementation

To implement the new report generation by integrating the existing reporting components with data science methodologies, changes to the data extraction, processing and visualization are required. The current report delivery methods to downstream and regulatory bodies, however, do not change. The end-users of the modified reports are internal stakeholders and internal management who mainly use reports for taking business decisions.

The following points should be considered during the implementation:

  • Existing reports can be used as input for the solution using Data Science methodologies. The raw data can be extracted from reporting tables via SQL. 
  • Once the data source is defined, the data required by the end-user needs to be collected.
  • The end-user report is generated using data processing techniques.
  • The end-user requirements on visualization and data representation must be considered.
  • Additionally, one needs to consider how the end-user analyses the data and how this process impacts his/her decision-making.
  • Based on this information, an alternative reporting solution is designed.
  • Along with the required data, visualization, suggested alternative decisions, summary etc. are delivered to the end-user as a report.
Key benefits

1. Reduced report processing time

  • At present, end users who receive reports in csv/xls/dat file formats, spend a lot of time on analysing data using Excel (like VLOOKUP, formulas etc). This can be significantly  reduced.
  • Automation helps processing more reports in less time.
  • Efficient methodologies help processing huge amounts of data which usually consumes considerable amount of time with manual processing.
  • Quicker decision-making is enabled as the key data points and alternative approaches have already been generated.

2. Improved report quality

  • Human error can be avoided with automated data analytics.
  • Reduced manual work and data entry results in less human error.
  • Better end-user satisfaction through avoiding mundane manual report analysis.
  • Reports, data and actionable insights are all in one place, which eliminates the need for end-users to spend a lot of time on analysis, so end-users from all business areas - traders, portfolio managers, marketing, finance, customer service, customers – can use data directly and reap benefits.

3. Enhanced data quality

  • Historic data comparisons and analysis help to gain more insights on trend analysis.
  • Reporting tools can fetch data from multiple sources (in addition to reporting database), e.g. from a transactional database which will make data analytics more robust.
  • With high quality data and reports, the error free decisions will help the business gain financial benefits at large.
Conclusion

Designing end-of-day end-user reports which automate processing of raw data and create visual aids to support internal stakeholder decision-making can dramatically reduce the extent of manual processing and thus decrease errors. Such process-oriented solutions can build on existing structures. They can be developed using design thinking methodology and bring substantial benefits in terms of speed and quality of information for decision making.

Capco has a longstanding history in the financial services industry. We understand what data is needed, where it comes from, who needs it and for what purposes, and we are experts at how to represent data visually. Contact us to find out more about data science methodologies for better, faster and more profitable business decision making. 

CONTACT

Andreas Pfeil, Senior Partner
T: +49 69 9760 9106
M: +49 172 165 3932
E: andreas.pfeil@capco.com