### Description Convert the existing 2 synthetic problems and 3 "real ones" to parquet For house prices - use the data split by time. Add the train/test as an additional dataset of historical prices.