> ## Documentation Index
> Fetch the complete documentation index at: https://docs.sourcemedium.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Modeling Philosophy

> How we help amplify your data capabilities

## Bootstrapping Data Teams

Our data modeling philosophy is grounded in the principle of abstracting away the complexity of raw data
transformation. The world of data can be a bad user experience, requiring endless hours of tedious work to produce usable datasets.
And, once you get to this point, the real value-add work hasn't even begun!

Our goal is to bootstrap engineering and analytics teams with high-quality, intuitive, and scalable
datasets so that they can focus on developing insights that differentiate their businesses instead of reinventing the wheel.

## How We Shape Our Data

Once we've ingested raw data from the source, we do the dirty work of [cleaning](/data-transformations/data-cleaning),
[enriching](/data-transformations/data-enrichment), and structuring our downstream models into several different shapes (see below) based on best practices to fit a variety of analytical use
cases. These include composable fact and dimension tables, tables purpose-built for specific visualizations, and datasets that
enable predictive analytics (and other advanced capabilities).

### Fact tables

Fact tables, indicated by the `fct` prefix, are central components of our schema that primarily store quantitative information
about events or transactions that have occurred. Each record in a `fct` table represents a unique event or transaction, characterized by
metrics or measurements such as sales amount or units sold. `fct` tables are immutable, meaning once a transaction or event is recorded, it
doesn't change, ensuring the integrity and historical accuracy of the data. These tables are often surrounded by dimension tables to provide context
to the facts through foreign keys that can be used for joins.

### Dimension tables

Dimension tables, indicated by the `dim` prefix, contain descriptive attributes related to the dimensions of the facts in `fct` tables.
These tables provide the context for metrics recorded in `fct` tables, such as time, geography, products, or customers. Unlike `fct` tables,
dimension tables can change. For example, if a product's description changes, the corresponding dimension table would be updated to reflect these
changes.

### One Big Tables

One Big Table, indicated by the `obt` prefix, is a comprehensive table that joins elements of both `fct` and `dim` tables into a single,
unified structure. These tables are designed to simplify the data model and represent SourceMedium's out-of-the-box semantic layer, which translates
the underlying data into concepts used by a business. OBTs are crafted to represent the business's underlying data in a way that's intuitive,
facilitating easier development of BI solutions by providing a consolidated view of data that might otherwise be spread across multiple `fct`
and `dim` tables.

### Report tables

Report tables, indicated by `rpt` prefix, are specialized tables designed for specific types of BI reporting. These tables are narrower in
scope compared to OBTs but may involve a higher degree of complexity, especially in terms of data aggregations, calculations, or transformations
required to support particular reporting needs. `rpt` tables typically serve specific reporting purposes, such as financial summaries, performance
dashboards, or operational reports, and are optimized for performance and clarity in those contexts.

### Summary tables

Summary tables, which include `summary` in the table name, are a variant of `rpt` tables that focus on providing aggregated views of
data, summarizing detailed information into higher-level insights. `summary` tables might aggregate data by time periods, geographic regions,
product categories, or other dimensions to offer concise, actionable information derived from more granular datasets.