Technical description


Introduction

Recommending companies in tenders has a great impact on Public Procurement. Search engine for Public Tenders is an innovative tool and very useful for companies (suppliers) which can search for suitable tenders for their profile. It also benefits the public procurement agencies which can automatically discover unknown companies at the moment. This platform uses a pioneering algorithm created ad hoc to search for compatible companies thanks to Machine Learning technology, in particular Random Forest Classifier.


Successfull case use

The proposed algorithm has been validated with remarkable results: the winning bidder is within the group of recommended companies in 31% of the tenders analyzed. In other words, the recommendation algorithm hits 31% correctly of the tenders. The study was supported by 102,087 Spanish tenders between 2014 and 2020 (public and free information) and business data from 1,353,213 Spanish companies obtained from the Spanish Business Registry. This successfull results have been published in a scientific journal, available in: Bidders Recommender for Public Procurement Auctions Using Machine Learning. You can read other academic articles related to Public Procurement carried out by us:
Spanish Public Procurement: legislation, open data source and extracting valuable information of procurement announcements.
Public Procurement Announcements in Spain: regulations, data analysis, and award price estimator Using Machine Learning.


Description of the algorithm to search for companies

1. Steps to train and create the machine learning model:

Figure_1_Bidders_Recommender_Algorithm

2. Steps to apply the created model on this website:

Figure_2_Bidders_Recommender_Application