@mention a person to add them as an attendee and they will be notified.
I. Opening
A. Greetings to attendees
B. Mr. Antley: LF AntiTrust statement.
II. Agenda - Data Modeling and Internal Data Work - Mr. Antley
Mr. Antley pulled up dialogue - graphic on earned premium - Greg W. spoke about car years last week
Mr. Antley: we want to discuss concept of business layer
Stat data as it is gives us what we need to do regulatory reporting. Specified accounting date and duration - clarification of coverages and exposures.
When we consider requests brought to us, some terms are brought up repeatedly: EP, incurred premium, exposures and premiums incurred and written, average expenditures and premiums, loss ratios, underwriting expenses, etc.
Today: we take input records and process those records and calculate all of these categories (various figures) on a quarterly basis. In current mode, grouped by policy and by quarter. Select queries feasible - yield year outputs. Calculations = aggregations.
In AWG: discussing what is the right level and how to set up HDS to make accessible to regulators without putting undue burden on rest of system
Question: we think of base layer from which we're answering queries, how should it be organized to make it accessible to those issuing queries?
Quarterly basis makes end report generation very easy.
Mr. Antley: how should data layer look?
Mr. Sayers: when looking at data layer, we must ask visibility to whom? Carriers e.g., want to see stat plans eventually with more data. Mr. Sayers: what are we asking carriers to provide and do they care about formats other than stat plans
Mr. Antley: carriers vs. regulators. For him, critical question: who is tasked with turning the stat records into a query that makes sense for regulators?
Mr. Sayers: we can start with stat plan as raw material, throw reports at it and ask how difficult is it to get from this stat plan data to report. And ask: Do we need an intermediate format?
Mr. Madison: If something is a raw element not included in current stat plan, must be added. Business layer is a black box for now. This meeting does input and output. Pointed to limiting focus of this meeting to data modeling alone. (Mr. Braswell agreed).
This meeting is not concerned with business layer: only concerned with how do we calculate earned premium.
Mr. Harris: we want to use existing stat plan as a means of building the plumbing. We want to be able to get from A to Z from an architectural perspective - request comes in, permission approved, extraction occurs, and then at EOD we have a report. Wants to go back to data model we have
Mr. Braswell: what are outputs that need to be clarified. Affirmed Mr. Harris's idea as prescient and valid.
Mr. Antley: having a difficult time transforming data
Mr. Harris: it works in AAIS why can it not work in idl, etc. Mr. Braswell: this is an architectural question.
Mr. Sayers: extraction pattern has been broken up into multiple levels to get it to work in AAIS. Getting it all to work in one extraction pattern overly complicated.
Mr. Madison: Most concerned with clarity of the rules: it is possible to over-engineer any model. His challenge: what are we trying to achieve. Separating business rules from shape of model of data.
Not a single straight move. Mr. Antley: confused about this. Summing exposure, premium, incurred claims/losses.
Time | Item | Who | Notes |
---|---|---|---|