The Table¶

Every table-shaped answer either library gives is a Table: a real data frame — pandas.DataFrame in Python, data.frame in R — that also carries what an actuary needs to trust it. Nothing about it stops being a data frame: slice it, merge it, plot it, feed it to any other package. The extras ride along.

Carried	What it is
`title`	one line naming the table
`basis`	one line of provenance — `Gompertz–Makeham (illustrative) · i = 4 %`
`notes`	the caveats, printed under the numbers every time
`provenance`	a content hash, timestamp, versions and call trail, once you `hard` it

Printing¶

Print a Table and the notes print under it. This is the house rule made mechanical: a number never travels without what produced it.

PythonR

>>> import scelo as sc
>>> sc.life_table().head(3)

> sc_life_table()[1:3, ]

   age        qx        px         lx         dx            Lx            Tx         ex
0   20  0.000250  0.999750  100000.00  24.963902  99987.518049  6.591309e+06  65.913088
1   21  0.000253  0.999747   99975.04  25.325403  99962.373399  6.491321e+06  64.929422
2   22  0.000257  0.999743   99949.71  25.732197  99936.844601  6.391359e+06  63.945747
— Life table · Gompertz–Makeham (illustrative)
  basis: Gompertz–Makeham (illustrative)
  · Mortality is Scelo's ILLUSTRATIVE Gompertz–Makeham basis (A = 0.00022, B = 2.7e-6,
    c = 1.124), not a published standard table: swap in your own qx column or parameters
    before relying on the figures.
  · Radix l(20) = 100,000; table closed at age 110 (qx set to 1). Lx uses the
    uniform-deaths approximation lx − ½dx.

In R, long tables truncate at 60 rows when printed; options(scelo.print_rows = 200) raises that.

Getting at the pieces¶

PythonR

t = sc.factors(i=0.04, n=10)
t.title          # 'Annuity & assurance factors · …'
t.basis          # 'Gompertz–Makeham (illustrative) · i = 4 %'
t.notes          # list of caveat strings
t.df             # the plain pandas.DataFrame, extras stripped

t <- sc_factors(i = 0.04, n = 10)
sc_title(t)      # "Annuity & assurance factors · …"
sc_basis(t)      # "Gompertz–Makeham (illustrative) · i = 4 %"
sc_notes(t)      # character vector of caveats
sc_df(t)         # the plain data.frame, extras stripped
sc_note(t, "reviewed 2026-08-23")   # append your own caveat

Subsetting keeps the extras: t[t$age < 65, ] in R (and slicing in Python) still knows its title, basis and notes. Stripping to the plain frame is always explicit — t.df / sc_df(t) — never accidental.

Rendering¶

t.to_markdown_report() (Python) / sc_markdown(t) (R) turn one Table into a Markdown block — title, basis line, pipe table, notes as bullets, and the provenance stamp once it is hard. report does the same for a whole pack.

Results that carry more than one table¶

Model runs return a small result object whose printable summary is a Table, with the richer pieces attached:

PythonR

m = sc.mack(sc.triangle(claims))
m.table       # the per-origin Table (latest, cdf, ultimate, ibnr, se, cv)
m.ibnr        # 1532963.3…
m.se          # 346036.4…
m.factors     # the link ratios
m.detail      # method internals (per-origin MSE, sigmas…)

m <- sc_mack(sc_triangle(claims))
m$table       # the per-origin Table
m$ibnr; m$se; m$factors; m$detail
summary(m)    # one-row summary frame

The same pattern holds for glm (coefficients Table plus fitted values, covariance, deviance), lee_carter (forecast Table plus ax, bx, kt, drift) and wmtr (year-by-year Table plus paths, outcome fractions, drivers).

Column names are part of the interface¶

The factor tables use the real actuarial glyphs — äx, A¹x:10, 10Ex — because that is what the quantities are called. In R, back-tick them: t$`äx`. In Python they are ordinary string labels: t["äx"].