Claims Loss Research Ideation Challenge

Claims Loss Research Ideation Challenge

Challenge Overview

Our client, an insurance company wants us to build a system that will crawl internal and external da sources and provide search results for the relevant claims.

 

The data is comprised of claims and policies data (clients buy policies and make claims that result in possible payout). Other data sources are internal company documents, and public data sources that should be determined (by you) in this challenge.

 

The ideal outcome of this challenge is a paper containing an analysis of the input data, list of identified trusted sources to crawl and how to connect the crawl results to claims data, and an architectural design of how a system would filter the data for a user query.

Background

Our client has a lot of large payouts in $250k-$2.5MM where a person deciding on a claim (adjustor) makes all the decisions without oversight and governance in place. Their job consists of analyzing the policy and the claim and then manually researching current historical precedent and current trends/news around industry and litigation of these claims.

Today, their best source of research is google.com or a paid for research forum – and the adjustor has to take the initiative to perform the research vs. being pushed content.  This is a blend of structured and unstructured data (i.e. take structured claims data and then crawl over public and trusted sources to bring together research / opinion of unstructured data) to help guide the claims adjustor’s strategy of settle the claim, litigation or mediation, reinsurance, and or closing the loop of communication with underwriting about how to go to market with future policy language.

 

The goal of Claims Loss Research is to:

• Identify trusted public sources of data that could be crawled - this includes news sites, paid publications, court filings (litigation results, court filings, etc)

• Provide details on how to connect the crawled data to the claims data

• Create filters to search the crawled data - to provide relevant search or categorical drill down overview through the data

 

Data Description

 

There are three data sources in this challenge:

1. Incident report data

2. Third party data - data sources identified in this challenge.

3. Policies and claims data

 

Incident report data contains the initial report on the insured asset - similar to a news report about a fire, flooding, etc. This report should be used to identify as much data about the insured party as possible - ex company name, location, products, industry category, etc. Goal is to use this info to search the relevant third party sources of data.

 

Third party data sources can include news reports, financial institution reports, etc and the goal is to find relevant documents that would help the adjustor get a better sense of what the appropriate next steps would be, and why (why is also important).

 

Policies-Claims data contains flat data records that:

• List the policy details (id, country, industry categorization, etc)

• List the policy exposure data - industry codes, amounts, addresses, risk scores

• List the claim exposure data - dates, amounts, claim status, causes of loss, etc

This data source is not considered a primary data source - at the time the initial report is submitted, there is still no connection between the company named in the report and the actual policy. This data source is provided mainly to provide insight into what additional info the adjustor might get from the internal company data. Additionally we only have access to the data definition document of the policies data, not the actual data. Check the sample and metadata file available in the forums for a complete definition of all data fields

Task Detail

In this challenge, the goal is to analyze the data, engineer the necessary features (if needed), identify the public data sources and build an algorithm/process to crawl and filter the data.

 

The main idea is to use the incident report data together with crawled data and to connect them using NLP - and we can rely on NLP service from Azure, Google, Amazon, Watson.

Here are some sample outputs that could be provided by the tool:

• Articles/documents/news related to a specific report (company, industry, type of loss, etc)

• Articles/documents/news about specific industries or hierarchical data with options to drill down into more specific categories - to provide info about latest trends, litigations, etc

• Information about the competing products (ex company that produces product X had a fire and can’t produce anything at the moment - are there substitute projects, are they available, etc)

 

The expected final deliverables are short white papers with data analysis (e.g., tables and figures to illustrate the analysis and recommendations), PoC codes (optional), and details on how to build the crawling and filtering in future challenges. We are not looking to build the entire analysis, crawling and filtering in this challenge - having the data analysis and detailed approach description that can be used in future challenges to actually build the system is what is expected in this challenge.

 

Submission Contents

A document with details for the data analysis and the proposed algorithm and/or a proof of concept solution, pseudo-code or any documentation/ previous research papers that helps illustrate proposal to create the final submission .

 

The final submission should be a report, more like a technical paper. It should include, but not limited to, the following contents. The client will judge the feasibility and the quality of your proposed approach.

1. Title : Title of your idea

2. Abstract / Description : High level overview / statement of your idea

   • Outline of your proposed approach

   • Outline of the approaches that you have considered and their pros and cons

   • Justify your final choice

3. Details :

   • Detailed description. You must provide details of each step and details of how it should be implemented

     • Description of the entire mechanism

     • The advantage of your idea - why it could be better than others

     • If your idea includes some theory or known papers;

       • Reason why you chose

       • Details on how it will be used

       • Reference to the papers of the theory

     • Reasonings behind the feasibility of your idea

4. Appendix(optional) :

   • Bibliography, A reference to the paper, etc.

Format

• A document should be a minimum of 3 pages in PDF / Word format to describe your ideas.

• It should be written in English.

• Leveraging charts, diagrams, and tables to explain your ideas is encouraged from a comprehensive perspective.

Judging Criteria

You will be judged on the quality of your ideas, the quality of your description of the ideas, and how much benefit it can provide to the client. The winner will be chosen by the most logical and convincing reasoning as to how and why the idea presented will meet the objective. Note that, this contest will be judged subjectively by the client and Topcoder.

Submission Guideline

You can submit at most TWO solutions but we encourage you to include your great solution and details as much as possible in a single submission.

Supplementary materials

You will be able to download from the Google Drive link posted in the forum.

Final Submission Guidelines

See above