About Data in Exabel

Exabel's product primarily deals with structured data in the form of time series. Most time series are linked to entities. For example, companies are a type of entity, and the company Amazon can have time series for its share price, revenue, web traffic visits, etc.

Entities allow us the flexibility to represent alternative data in its full breadth and depth, catering to how each data set can provide insights at various levels of detail. For example, a card transaction data set will typically have time series data at brand/merchant-level, which we can represent by creating a brand entity for each brand in the data set, along with various time series (e.g. spend, transactions, accounts) linked on each brand. An app traffic data set could have data at app-level as well as app-country, with DAUs, MAUs and sessions time series. This flexibility extends to your own proprietary data as well.

FactSet's entity master, containing 10 million public & private companies is used as a scaffold against which all data sets are mapped. This allows for the harmonization of data across vendors, who often use different ticker/company identifiers; and also allows for easy browsing and overlays of data across sources.

These entities are linked via relationships, which allow for browsing/searching for entities (e.g. show me the apps that are owned by Amazon) and for aggregating time series data at scale (e.g. find all the apps owned by Amazon, and sum their DAU time series).

Organizing data into types & data sets

Each of these building blocks (entities, time series, relationships) have higher-level groupings that allow users to more easily find them.

• Entities are grouped into entity types. For example, company entities are grouped into the company entity type, while vendor A's app entities are grouped into the app entity type. When searching for entities, this makes it possible to narrow down what you are looking for.
• Time series are grouped into signals, where every time series in a signal has the same meaning across entities. For example, all of the DAU time series in an app traffic data set can be grouped into a DAU signal. This allows you to generalize analysis and visualization - e.g. by KPI mapping the DAU signal to total revenue for all companies in the app data set, or by creating a chart template that will plot the DAU signal for any app that you select to visualize.
• Relationships are grouped into relationship types. For example, every company → brand relationship in a data set can be grouped into a HAS_BRAND relationship type.

Finally, this data (entities, time series, relationships) can be packaged into data sets. This helps you to more easily browse for specific data, understand the source of the data you use. It is also the means by which vendors control data entitlements.

Each data set is defined simply as a set of signals. For example, a simple card transaction data set may be defined as 3 signals - spend, transactions, accounts. From here, the Exabel system identifies the entities (e.g. brands) that have those signals' time series, and includes these as part of the data set. Importantly, their parent entities are also included as part of the data set. For example, if the card transaction data set only had time series at brand-level, but the brands were connected to companies via relationships (ownership relationships, to be precise), the parent companies would also be included in the data set.

Namespaces & Data Separation

Exabel uses the concept of "namespaces" to keep customer data segregated and private to users of that customer.

Each vendor's data is loaded into their own namespace. Buy-side customers who subscribe to a vendor data set will be able to access the data for that data set in that vendor namespace. (If a vendor has data sets A and B, but a customer subscribes only to A, they will only be able to retrieve time series data for A.)

There are a set of entities in the global namespace - mainly the FactSet entity master, and a set of country entities. All vendors & customers can connect their data to these global entities.

If you have an Exabel Enterprise license (needed for you to import data), you will have your own namespace, eg customer1. Data imported by you will be created in that namespace and therefore private; this includes entities, relationships, signals and time series.

Glossary

• Namespace: allows for the separation of data between vendors & customers.
• Data set: a collection of signals, such as price and revenue, those signals’ time series and all entities that have those time series as well as their parent entities.
• Signal: a group of time series that have the same meaning across multiple entities. For instance, Close_Price is a signal of daily closing share prices for all listed company entities.
• Time series: a series of data points each with corresponding timestamps. Every time series belongs to a signal, linked to the entity they correspond to. For instance, the time series of Apple daily closing share prices belongs to the Close_Price signal.
• Entity: represents real-world entities, such as companies, brands, apps, etc. For instance, the company Amazon, Inc. is an entity.
• Entity type: a group of entities having similar real-world meaning. For instance, company is an entity type. Every entity has a single entity type.
• Relationship: connect entities with each other. For instance, the company entity Volkswagen and the brand entity Audi may be connected with a relationship of the type HAS_BRAND.
• Relationship type: a group of relationships having similar meaning. For instance, HAS_BRAND describes all relationships between a company and the brands it owns.