Formula

Formula features are used when we need to create features on top of other features of the same entity. It is commonly used to enrich an entity with more complex calculations that require multiple aggregations or calculations to be done first.

Simple formula feature

In this example, we define simple metric features for the entity customer:

# customer.yml
features:
  
- type: formula
  name: is_active_customer
  data_type: boolean
  sql: IFF(({last_login_date} > date_add('day', -30, current_date())) and {total_orders} > 100, true, false)

On this example, we already have two features on the customer level: last_login_date and total_orders. We use these two features to define which customers are active customers.

formula is a special type of feature, which is not reliant on related data assets or related entities like other feature types, as it creates features based on features of the same entity.

type

The feature type. In case of formula features, it should be set to formula.

name

Give the feature a name.

data_type [optional]

Specify the feature data type. If no data_type specified, Lynk will assume the data type is string.

The options for data types are:

string

For any type of string data type

number

For any type of number data type. For example: integer, float, decimal etc..

bool

For boolean data type.

datetime

For any type of time-based data type. For example: date, timestamp, datetime etc..

sql

The formula definition.

Any SQL code applies here as long as:

1. It is based on features already defined on the entity

2. it does not have aggregate functions

Referencing a feature in the SQL is done by using {FEATURE_NAME}

In case you have an aggregate function in a formula, you probably need to create another metric feature and then create the formula feature on top of it

---

Understanding Formula Features

In case we need to create a new feature which is composed by other features of an Entity, we can use the Formula feature type.

Unlike field, metric and first-last features, which are built by enriching an entity from other data assets or other entities, formula features are meant for enriching an entity from features of that same entity.

To better understand how Formula features work, look the following diagram;

Formula Feature Diagram

The above diagram shows an example when a customer (entity) already has two existing features last_order_date and total_orders. We can create a new Formula feature is_active_customer as a composition of the existing features last_order_date and total_orders.