Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.thedatacity.com/llms.txt

Use this file to discover all available pages before exploring further.

​
Why we update RTICs

Annually, our Real-Time Industrial Classifications (RTICs) undergo updates to verify company status (dissolved or unavailable companies), ensure website matches, capture or add new companies, incorporate new keywords, and improve data accuracy and relevance. image png Dec 20 2023 02 41 05 3364 PM The above image shows RTIC updates over the past few years - updated 20th December 2023 Types of RTIC update Our RTICs undergo two types of updates to ensure accuracy & relevance
  1. Maintenance Update: This is a monthly automated update that ensures we continually classify new companies into RTICs where required.
  2. Annual Update: This is an annual update focused on substantial changes such as changes in the taxonomy, updating the underlying training set & retraining the Machine Learning list.
How is a Maintenance Update different from an Annual Update?
  • Maintenance Update: Focuses on applying existing classifiers to fresh company data.
  • Annual Update: Involves more substantial changes, including revising the taxonomy, retraining Machine Learning models, updating training sets, and adding new verticals if required.
What is a Maintenance Update? It is important that our RTICs provide the truest representation of a sector possible. To achieve this, Maintenance Updates run monthly and ensure that we continually classify companies with matched websites into RTICs. At present we account for:
  • Companies newly incorporated with a matched website 6 months ago, allowing time for the site to develop so we can accurately assign an RTIC.
  • Companies newly incorporated within the last 6 months with a manually matched website.
  • Existing companies with updated or newly reported websites since the last maintenance update.
This process will be subject to further improvements and changes in the future. Maintenance Update Process: Monthly Re-creation of Classifiers:  Every month, we re-create the existing RTIC classifiers using the original training sets and the websites/web text associated with companies as of the last Annual RTIC update or the RTIC creation date, if it was created less than 12 months ago. Consistency with the Last Annual Update:  
The existing classifiers (as described above) are used to ensure that the same relevant terms and patterns identified during the last Annual Update or the initial creation of the RTIC for newer ones are consistently applied. This approach ensures that companies added during the Monthly maintenance updates are classified using the same criteria as those already in the RTIC, maintaining a consistent definition across all updates.” Example:
  • RTIC0052| Fintech was originally created on 16-09-2021
  • It was last Annually Updated on 09-10-2024.
  • A Maintenance Update is scheduled for June 2025.
  • The classifier used for this Maintenance Update was created in 2024, not 2025, to retain consistency in the classification applied to the existing and newly classified companies
Maintenance Update QA: Maintenance Updates also undergo a QA process.
  • This is to check for the possibility of false positives and verify the % increase of company additions to an RTIC. A large % increase could indicate an issue with the underlying training set.
The following checks are performed:
  • Manually verify that a company has been correctly classified into the correct RTIC sector by checking its matched website.
  • Manually verify that a company has been correctly classified into the correct RTIC vertical by checking its matched website .
Only companies that are correctly matched to both their sector and vertical are approved to be pushed live to the platform. During the maintenance QA process the following data is logged:
  • Total amount of companies considered
  • Of which X were newly incorporated 6 months
  • Of which X were manually reported website matches
  • How many unique companies were classified
  • How many additions were classified in total to RTICs
  • Of those additions how many were approved after QA for each RTIC
This information is logged and tracked internally to monitor changes to RTIC composition over time. What is an Annual Update? Annually, our Real-Time Industrial Classifications (RTICs) undergo an Annual Update process. We prioritize updating RTICs based on how often they’re used on the platform and the date of their annual update Our Annual Update includes the following: Review of the Taxonomy: Revisit the Taxonomy of the verticals within the RTIC to make sure we capture the evolving nature of the sector. This could include building new verticals or updating the existing verticals. Review/Update the training set: We update our training set by replacing inactive companies with similar companies, removing those we can’t replace, and adding new companies that reflect current evolution of the sector List Rebuild & Quality Assurance To improve the accuracy and relevance of the list after updating the training sets, we re-run the list and perform a thorough quality assurance check to catch any false positives. Additionally, we manually review a sample of 30 companies to ensure they align with the RTIC sector and vertical. Updated RTICs Once the update process is complete, we release the updated RTIC to our platform, as shown in the image below. Newly updated RTIC have “revised” labels on them, this then becomes the Revised date for an RTIC. image png Dec 20 2023 02 55 22 0873 PM