Reinsurance Optimization with Markov Chains

Preprocessing

Make sure the file has proper headings, if not add titles:

Make sure all of the perils are accounted for

Run the script to turn raw YELT data into normalized data where we only keep the columns we want and transform all values into small (as possible) integers and floats.

cd preprocessing/
python normalize.py --events="../data/raw_yelt/YELT_RB25_Reinsurance_GROSS_USD_15052025_titles.csv"

Takes this normalized data and puts it into a sqlite3 DB, doing preprocessing of the different stop possible to choose from given risk constraints.

Building a SQL Database

python to_database.py --events="../notebook/data/YELT_RB25_Reinsurance_GROSS_USD_15052025_titles-normalized.hd5"

Run the Markov Chain

cd ../optimization/
python run.py --seed 42

Select a Subset for Visualisation

cd ../visualization/public/
python select.py