Learn 07~9 min

Reading model names

Most readers encounter a model name before they encounter its SQL. Reading the name helps you choose what to inspect next.

A model name is a small sentence

Names in this project normally contain three parts: a layer prefix, the domain subject, and a suffix where the model's shape needs to be made explicit.

int_hba1c_latest

modelling block · HbA1c results · the most recent result per person

fct_person_diabetes_register

reporting fact · a person's diabetes state · a disease register

stg_olids_observation

staging model · the OLIDS observation table · universally cleaned

fct_person_sus_uec_recent

reporting fact · a person's urgent and emergency care activity from SUS · a recent time window

Part	Question it answers	Examples
Prefix	What role does this model play?	`stg_`, `int_`, `fct_`
Subject	Which domain concept is it about?	`person_diabetes`, `blood_pressure`
Suffix	Which shape or variant does it contain?	`_all`, `_latest`, `_register`

Not every name needs all three parts. A staging model mirrors its source, so stg_olids_observation is complete. A canonical dimension such as dim_person needs no suffix.

The grammar spans every dataset the project models, not just OLIDS primary-care records. The same three parts name SUS hospital activity (fct_person_sus_apc_recent), GP appointments (fct_person_gp_recent), waiting-list pathways (int_wl_current) and resource use (fct_person_resource_index). Learning to read it once makes every source's models legible.

Prefix: the model's role

Prefix	What to expect
`raw_`	A generated one-to-one interface to a landed source table
`stg_`	One source made legible and ready for downstream use
`int_`	A purposeful transformation that prepares data for marts
`dim_`	Descriptive context used to understand and group facts
`fct_`	An event or measurable state at a declared grain
`pit_`	A point-in-time view for retrospective reporting
`obt_`	A wide analytical table composed from established concepts
`dq_`	A data-quality output containing records that need attention

The prefix follows the model's responsibility, not its size or its grain.

The analytical tables chapter explains facts and dimensions through examples. Here, read the prefix as a clue to the intended role. It does not prove the grain or tell you whether the model fits your population.

Suffix: the model's shape

A suffix is useful when it distinguishes models about the same subject or warns that one row means something different.

Suffix	What it tells the reader	Example
`_all`	Every qualifying event; several rows per person are possible	`int_hba1c_all`
`_latest`	The most recent qualifying event at the model's documented grain	`int_hba1c_latest`
`_current`	The state effective now	`dim_person_current_practice`
`_historical`	The history of a changing state	`dim_person_demographics_historical`
`_register`	A defined disease-register population	`fct_person_diabetes_register`
`_summary`	A roll-up at the grain named elsewhere in the model	`fct_ltc_lcs_practice_summary`

Joining int_hba1c_all to a person-grain model can multiply its rows. The _all/_latest pair makes that grain difference visible before the join is written.

Related names make differences easier to spot. Compare a history model with its latest-result model, then check their descriptions to see which date, population and selection rule each uses.

The repository already contains dozens of person dimensions, latest-event models and person-register facts. Looking across one family is often the fastest way to understand its taxonomy because the repeated parts stay fixed while the meaningful differences become visible.

Search	Models you will meet	What the family teaches
`dim_person_`	`dim_person_age`, `dim_person_ethnicity`, `dim_person_current_practice`, `dim_person_demographics`	The person is the described entity; the remaining words identify the attribute, relationship or composed person view.
`_latest`	`int_hba1c_latest`, `int_blood_pressure_latest`, `int_bmi_latest`, `int_smoking_status_latest`	Each model selects one qualifying record from a richer history. Compare it with the matching `_all` model to see the grain change, selected date and tie-breaking rule.
`fct_person_` + `_register`	`fct_person_asthma_register`, `fct_person_ckd_register`, `fct_person_diabetes_register`	These are related person-grain clinical facts. Compare their population, reference time, evidence and resolution rules rather than assuming every register is constructed identically.
`fct_person_sus_`	`fct_person_sus_uec_recent`, `fct_person_sus_apc_recent`, `fct_person_sus_op_recent`	One person-grain activity summary per SUS dataset: urgent and emergency care, admitted patient care, outpatients. The shared shape makes the differences worth reading: each dataset has its own attendance, admission and status rules.

Try reading the names

Use this exercise to practise the grammar. Finding a plausible name is the first step; the next chapter shows how to inspect descriptions, tests and lineage before deciding to use it.

Find the model · 1/5

Your analysis needs each person's most recent blood pressure reading.

399 real models from the project · prefix colour = layer

Naming a model of your own

Choose the name after deciding what the model does. Use the same subject vocabulary as neighbouring models. Add a suffix when it distinguishes a time, grain or selection rule; words such as _final and _new tell the next reader very little.

A shared pathway model should describe pathways. A published model can appropriately name its consumer, such as asthma_dashboard_base, because serving that product is its purpose. The naming reference keeps the patterns together for lookup. Finding models now puts the names to work in a complete search.

Read the taxonomy

0/2 answered

1What is the important difference between int_hba1c_all and int_hba1c_latest?
2You think a new person-level housebound model is needed. What should happen first?