andras/perfect-postcode
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages 2 Wiki Activity Actions

Small fixes

Browse source
This commit is contained in:
Andras Schmelczer 2026-06-14 14:52:44 +01:00
parent 54fbcb1ea6
commit 083f8a982e
24 changed files with 1505 additions and 79 deletions
Download patch file Download diff file Expand all files Collapse all files

20
pipeline/transform/join_epc_pp.py
View file

@ -52,6 +52,13 @@ JUMP_MIN_PRICE = 2_000_000
MIN_BUILD_YEAR = 1700
MAX_BUILD_YEAR = 2030
# Open-ended "before YYYY" bands (EPC only emits "before 1900") have no lower
# bound. Rather than dropping these mostly Victorian-and-older dwellings, we
# publish a representative year this many years under the stated upper bound
# ("before 1900" -> 1890) so they still carry a build year and survive an
# age-range filter; is_construction_date_approximate flags it as an estimate.
OPEN_BAND_YEARS_BEFORE_BOUND = 10
# Plausibility bounds for raw EPC dimensions. EPC lodgements contain data-entry
# errors (0 m storey heights, 116 m "interior height", 9,210 m² floor areas, 99
# habitable rooms) that otherwise propagate verbatim into the published per-
@ -68,10 +75,13 @@ def epc_band_to_year(band: pl.Expr) -> pl.Expr:
EPC age bands are ranges (e.g. ``1950-1966``); we use the band MIDPOINT
(1958) rather than the lower bound, which previously biased every band-derived
year ~10-15 years too young. Open-ended lower bands (``before 1900``) are too
wide to pin to a year and return null. Single-year / ``... onwards`` bands use
that year. Already-numeric inputs (a year produced by an earlier call) pass
through unchanged. Years outside [MIN_BUILD_YEAR, MAX_BUILD_YEAR] are nulled.
year ~10-15 years too young. Open-ended lower bands (``before 1900``) have no
lower bound, so we publish a representative year OPEN_BAND_YEARS_BEFORE_BOUND
years under the stated upper bound (``before 1900`` -> 1890) rather than
dropping the dwelling; is_construction_date_approximate flags it as an
estimate. Single-year / ``... onwards`` bands use that year. Already-numeric
inputs (a year produced by an earlier call) pass through unchanged. Years
outside [MIN_BUILD_YEAR, MAX_BUILD_YEAR] are nulled.
"""
text = (
band.cast(pl.Utf8)
@ -82,7 +92,7 @@ def epc_band_to_year(band: pl.Expr) -> pl.Expr:
high = text.str.extract(r"(\d{4})\D+(\d{4})", 2).cast(pl.Int32, strict=False)
year = (
pl.when(text.str.starts_with("before "))
.then(None)
.then(low - OPEN_BAND_YEARS_BEFORE_BOUND)
.when(high.is_not_null())
.then(((low + high) / 2).round(0).cast(pl.Int32))
.otherwise(low)