Informationen zur Anzeige:

Masterand Quantitative Modelling & Structuring (w/m/d)
Aktualität: 08.03.2024

Anzeigeninhalt:

08.03.2024, EnBW Energie Baden-Württemberg AG

Einleitung

At EnBW Trading, many models are running for a large variety of purposes: forecast of renewables production and prices (power, fuels, CO2), live curves estimation based on of the products traded on the market, etc.These models take inputs as timeseries, which come from many different sources such as external data providers.
It is frequent that these input timeseries are incomplete: there can be NaN (not a number) or even missing data (jump in the data timestamps). These inputs of poor quality are deleterious for the overall forecast precision and availability. There is a strong need for high-fidelity data imputation.Recently, GAN (Generative Adversarial Nets) models have demonstrated outstanding abilities to generate realistic data under various forms (images, videos, sounds, time series) and with severe constraints.The goal of this master thesis is develop a GAN model, and to apply it in business related use-cases where the need for data imputation has been identified.
EnBW Energie Baden-Württemberg AG
Energie & Umwelt

Masterand Quantitative Modelling & Structuring (w/m/d)

Abschlussarbeiten, IT

Unser Angebot

Spannende und anspruchsvolle Aufgaben eröffnen vielfältige Perspektiven zur persönlichen Entwicklung
Kollegiale Zusammenarbeit im Team und moderne Arbeitsplätze zeichnen uns aus
Vielfältige Events bieten Austausch- und Netzwerkoptionen für unsere Studierenden im Rahmen des Studi-Energy-Programms
Flexible Arbeitszeitmodelle ermöglichen die Vereinbarkeit von Job und Studium
Attraktive Vergütungen eröffnen Möglichkeiten: 1.936,- EUR mtl. brutto bei allen Praktika und Abschlussarbeiten und ab 16,- EUR je Stunde brutto bei Werkstudierendentätigkeiten im Bachelorstudium bzw. ab 18,- EUR brutto im Masterstudium

Vielfältige Aufgaben

Literature survey on ML (especially GANs) for data imputation + presentation
Thorough analysis of the incomplete timeseries to cure
Model development and Testing
Application to different use-cases
Write Master thesis

Überzeugendes Profil

You are a student in any scientific fields (data-science, informatic, mechanics, aerospace, economics, electronics, chemistry) with a strong background in Machine Learning (university projects, passed exams, side-projects, HiWi, etc)
You have a special taste for coding and model development and high skills in all Python librairies revolving around the ML universe (pandas, jupyter, tensor-flow)
You are not afraid of diving into the details of neural networks and have a good understanding of the maths behind, in terms of operation and data structure