Darts – Time Series forecasting

Customer

open source by Unit8

Problem

  • Any quantity varying over time can be represented as a time series: sales numbers, rainfall, stock prices, CO2 emissions, Internet clicks, network traffic, etc. Time series forecasting — the ability to predict the future evolution of a time series— is a key capability in many domains where anticipation is important. Although there exist many models and tools for time series, they are often non-trivial to work with, because they each have their own intricacies and cannot always be used in the same way

Solution

  • Darts is our open source library for time series forecasting, attempting to simplify time series processing and forecasting in Python

Impact

  • Speed-up the process related to time series forecasting in order to
    – Decrease costs
    – Improve accuracy
    – Reduce manual work
Read more

Reporting Automation

Customer

Swiss Insurance company

Creating automatic reporting tools through data centralisation, eliminating tedious and repetitive tasks anc increasing quality of reports

Challenge

  • An insurance company creates quarterly financial reports. They were manually extracting data, exchanging it via emails back and forth for aggregation in Excel and further validation

Solution

  • Ingestion of the source data to the data lake and implementation of the pipeline to validate and clean the data.
  • UI application was built to streamline the process reporting process by giving access to all parties and allowing modifications if necessary

Impact

  • Eliminated the repetitive and tedious manual process, reducing the amount of time needed to create reports
  • Improved the quality of reports by reducing the margin of human error

Ontology normalisation

Customer

Swiss Insurance company

Normalising existing ontology and migrating to a new standardised ontology version

Challenge

  • Existing Ontology too large, makes it hard to manage and analyse

Solution

  • Normalised existing tables by splitting them into multiple units
  • Built toolkits and ontology library to support development of new ontologies
  • Conventions and development guidelines
  • Performance optimisation and reporting

Impact

  • Normalised 3 tables resulting in final ontology of 18 tables (approx. 2.3B rows)
  • 30% reduction in total ontology build time
  • 60% reduction in disk spillage

Live Contract Renewal Dashboard

Customer

Swiss Insurance company

Creating a live contract renewal dashboard allowing underwriters and region directors to have a better understanding of trends and current contract status

Challenge

  • Create live dashboard and weekly report to track contract renewal progress

Solution

  • Built a big data pipeline to track contract status progress and calculate year-to-year metrics
  • Developed application to visualise live data
  • Created charts enabling easy visualisation of contract renewal status within the market

Impact

  • Underwriters are now able to see live data and adjust contract pricing
  • High-level automated reports allow region directors to have a better understanding of current trends and statuses of contracts

Analytics Maturity Assessment

Customer

Swiss Insurance company

Conducting a comprehensive maturity assessment of analytics capabilities, allowing for better future strategic planning and increased transparency

Challenge

  • A global insurance company was facing different levels of analytics maturity across its business units and had little visibility on the existing capabilities, tools, standards and project roadmaps.

Solution

  • Co-developed and conducted a comprehensive analytics maturity assessment by means of a written questionnaire and follow-up interviews.
  • Many dimensions were considered, including analytics strategy, design and ethics.

Impact

  • Substantial gains in transparency for the client on its existing capabilities.
  • Results served as a baseline for future group-wide strategy initiatives regarding analytics, such as strategic workforce planning or analytics governance.

Analytics Governance Framework

Customer

Swiss Insurance company

Developing a comprehensive governance framework combining findings from a maturity assessment, ISO standards and sector best practices

Challenge

  • No existing framework defining the ideal state and goal of the advanced analytics governance efforts

Solution

  • Co-designed a comprehensive ideal governance framework that set a target for the governance and analytics efforts in the organisation
  • Framework developed based on previous maturity assessment, ISO standards and external best practices

Impact

  • Mitigation of operational, financial, legal & reputational risks from analytics solutions through best-in class governance practices
  • Standardised and more efficient development practices

Analytics Center of Excellence

Customer

Swiss Insurance company

Supporting a strategic initiative to migrate current pipelines and use cases to the Foundry platform

Challenge

  • Need to offer company-wide support in the transition from old systems to Foundry
  • Multiple teams without prior Foundry knowledge needed to move their pipeline and use cases on the platform

Solution

  • Multiple support processes depending on the needs
  • Support including: office hours, architecture reviews, pair programming, implementation & handover, code reviews, operational support, etc.

Impact

  • Supported strategic migration to Foundry, onboarding new users and setting up data foundations
  • Platform usage increased from 2’000 to 5’000 users
  • Provided support for 25+ projects
  • CoE team grew from 2 to 10+ people in 2 years

Exposure analytics

Customer

Leading Swiss insurance company

Creating an application to present a heatmap of insured value allowing to easily spot overexposed areas and limit financial losses due to natural disasters

Challenge

  • Existing struggle to identify overexposed areas, either in terms of total insured value or per given peril

Solution

  • Creation of an application to present a heatmap of insured value, either total or per peril, for all granularity levels (country, region, etc.)
  • Integrated many filtering options allowing for deeper analysis in certain areas if necessary

Impact

  • Client is able to quickly spot overexposed areas and coordinate with underwriters to either stop offering insurance for items in this area, or change the pricing
  • Lowering of financial risk in the event of a natural disaster

Claim risk scoring

Customer

Leading Swiss insurance company

Detecting potential claim fraud among health insurance clients

Challenge

  • Undetected fraudulent insurance claims may cause a significant blow to insurance companies’ revenues

Solution

  • Implemented fraud scoring KPI’s together with visualisation
  • Enabled multiple insurance companies to use the API to validate claims of their customers
  • Ensured auditability and idempotency of the results

Impact

  • Automated the fraud-flagging process
  • Created tools allowing enhanced manual inspection of the most suspicious claims for multiple companies

Climate risk scoring

Customer

Swiss Insurance company

Developing a climate risk scoring model scalable to the entire insurance portfolio and able to determine climate risk exposure in real time for any locations

Challenge

  • Current climate risk scoring model is limited
  • Scoring too time-consuming and inability to score for many locations at once
  • Model needs to be scaled to the whole insurance portfolio

Solution

  • Implemented new model in Big Data environment capable of ingesting high volumes of data
  • Used external weather model to predict climate risk score and estimate impact on insurance portfolio

Impact

  • High performance model now allowing to evaluate climate risk scores in real time
  • Ability to display climate risk scores for any of the more than 2B locations in the insurance portfolio

Financial health watchlist

Customer

Swiss Financial Institution

Developing a centralised solvency and risk assessment model to create risk exposure reports automatically and improve data quality

Challenge

  • Difficulty to predict total exposure to financial losses resulting from companies at risk of becoming insolvent
  • Current risk reports are generated manually and data quality is not guaranteed

Solution

  • Central gathering and processing of the relevant partner data allowing for central risk modelling and data quality checks
  • Automatic creation of monthly risk reports based on the curated data

Impact

  • Improved risk management capabilities through information centralisation and standardisation
  • Improvement in data quality and report generation
  • Possibility to create reports dynamically to visualise more aspects than before

Previous

Data

Next

Finance
close

This page is only available in english