Desk Review Spreadsheet Modifications

[Damonamajor]

2026-02-05-footloose-rob is the model run which relates to the stage 2 changes.

Two other oddities were discovered when looking at tieback cycles. First, there are some pins which have NA tiebacks and are part of a hypothetical tieback cycle.

SELECT *
from default.vw_card_res_char
where tieback_key_pin = '07212150060000'
or pin = '07212150060000'

And there are pins which go to tiebacks which do not have a valid pin

SELECT *
from default.vw_card_res_char
where tieback_key_pin = '01011260410000'
or pin = '01011260410000'

It also removes all references to comps from the desk review spreadsheets.

Completes the following requested tasks.

• Rename the flag_proration_sum_not_1 column to "Proration Rates Don't Sum to 100%" in the Excel template.
• Add new flag for proration tieback cycles
• Per feedback from Valuations, let's replace these columns with one column/section "HomeVal Report". This column should contain a link whose display text is the PIN for the row, and whose link target is the staging HomeVal report for that PIN in 2026.
• Let's add two new columns to the "Sales Information" section (blue background), one for each sale (This was modified to be the existing sale_recent_1_outlier_reason.
• Row sorting is not always intuitive. We should sort the rows based on the following logic that looks at the "row label:
  First, sort by neighborhood number ascending
  Within each neighborhood, sort by class code ascending
• Column values should be reported in units of MV (market value), not AV (assessed value)

[Damonamajor]

This join actually results in slightly less TRUE observations since the proration cycles count PINs which don't exist as being in a tieback cycle. See the example in the PR comment for an example.

[Damonamajor]

There are no duplicates for pin at the moment, but would we want to add a distinct at the end just to make sure it stays that way?

[Damonamajor]

we don't have the comp sheet in the desk review spreadsheet anymore.

[Damonamajor]

switch to market rather than assessed values

[Damonamajor]

Add igraph to the depends section of the DESCRIPTION. It seems to already be installed in renv, but let me know if not.

[Damonamajor]

Do we want to expand this from tieback cycles to also include irregular tieback pins? At the moment we also reference tiebacks which don't exist as an actual PIN, and NA tiebacks. The graph treats NA's as a unique observation so it groups it as >1. I'm pretty sure we can filter NA's out to make this strictly tieback cycles, but let me know what you want.