IDSEM, an invoices database of the Spanish electricity market.

ABSTRACT: This article describes a new database of electricity bills related to energy consumption in Spanish households. The dataset includes individual invoices containing information about the consumption and billing of each supply point. These documents include additional data about the customer, the contract, and the electricity company. We propose a pipeline for the creation of bill contents through a simulation process based on regulations and statistics from official bodies and electricity companies. This makes it possible to generate many documents with synthetic data. The simulation is based on 86 different labels, which are necessary to create realistic invoices. The dataset has 75 000 documents in PDF format with their corresponding labels in JSON files. It is useful for training machine learning algorithms and, in particular, for developing methods to automatically extract information from the bills. It is also interesting to design new algorithms for analyzing the behavior of electricity markets from different perspectives.

SUBMITTER: Sanchez J 

PROVIDER: S-EPMC9809319 | biostudies-literature | 2022 Dec

REPOSITORIES: biostudies-literature