data-raw/README.md

In jimbrig/lossrx: Actuarial Loss Development and Reserving with R

Data Preparation Directory - data-raw

Folders

➜ tree data-raw
├───cache
├───database
│   ├───CSV
│   └───SQL
├───original
│   └───downloaded
│   └───mockaroo
├───scripts
└───working
    ├───exposures
    ├───industry
    ├───lossruns
    ├───policies
    └───priors

Cache

A cache directory for intermediate caching of long-running data manipulations.

Database

This directory represents the data preparation portion of the final inst/database folder which houses the final utilized database files.

In particular this folder should have:

• CSVs for each table (see database/CSV)
• SQL table creation files for each table (see database/SQL)
• Any other auxillary or referenced workbooks or data files that are implemented into the database
• A final db.sql SQL script that generates and seeds the database.

Scripts

The scripts directory houses all the data preparation R scripts.

Currently the following scripts are used:

• db_init.R: This script generates the database against a running PostgreSQL database container image.
• dataprep.R: This script downloads, extracts, and transforms data that is included, documented, and exported with the package as .rda files in the data directory. Specifically, it generates the losses and exposures data frames.
• claims_transactional.R: This script simulates an example transactional claims dataset and is exported with the package as claims_transactional.

Any other scripts are extras and may be ignored.

Original

The original directory houses raw, original data downloaded or brought in from an external source. This data is assumed to be as raw as possible.

• downloaded - houses data downloaded via ropensci's piggyback package.
• mockaroo - Mocked data frames created using the Mockaroo API.

Working

The working directory houses data that has been manipulated from the files in the original directory. For example, extracted zips, renamed files, re-formatted tables, etc. The working data is what is actually used when preparing the datasets for use in the package.

Data Lineage: ORIGINAL --> WORKING --> USED

Lossruns

Lossruns to be scrubbed.

Exposures

Exposure Data for various scenarios and coverages.

Industry Data

• Loss Development Factors (LDFs)
• Increased Limit Factors (ILFs)
• State filed loss-rates (Loss Costs)
• Trend Factors
• Decay Factors
• Ultimates or distributional parameters

Prior Working Data

• Prior data to be used as if prior analysis have been performed in order to aid the Actual vs. Expected workflow.

jimbrig/lossrx documentation built on Dec. 20, 2024, 3:46 a.m.