Project 7: Building S&P500 portfolios with decreasing carbon footprints (2022)
In this project, we analyze different asset management strategies as well as KPIs to compare them. The project simulates investment strategies throughout the duration of 20 years with models such as Equally Weighted, Global Minimum Variance, Max Decorrelation, Max Diversification Maximum Sharpe Ratio. The assets are initially selected by taking into account the 5 best assets with the lowest carbon footprint from each of the 10 sectors of the S&P500. For each optimization strategy, the objective functions are penalized by each asset’s carbon footprint. The result are then analyzed with KPIs including : Return, Volatility, Information Ratio, Diversification Index, Sharpe Ratio, Tracking Error, Excess Return. The entire project was developed from scratch and conducted in French. The report notebook can be found in the code folder under the name of RAPPORT_DIA_MILADINOVA_PROJET_FINANCE.ipynb This project was conducted during the Msc. in Energy Systems Optimization of Mines Paris in collaboration with Ms. Simona MILADINOVA and Mr. Seydou DIA under the supervising of Mr. MARTELLINI, Head of EDHEC Risk-Institute, as part of the Sustainable Finance course.
Project 6: Modelling electricity markets : Study of the benefits of interconnections on prices (2022)
if link doesn’t work, click HERE to access project
In this work we study an optimization model of electricity coupled markets that maximizes welfare. The goal of the project is to analyze the impact of interconnections on electricity prices. We first conduct the study with 3 price zones each containing 100 offers and buyers. We then extend the model to a 35 zones problem. The two indicators analyzed are congestion rent and price convergence. This project was conducted during the Msc. in Energy Systems Optimization of Mines ParisTech in collaboration with Ms. Simona MILADINOVA and under the supervising of Mr. Jean-Paul MARMORAT, research director at Mines ParisTech.
- Slides IN FRENCH
Slides contain dynamic plots. Please download them in pptx otherwise you won’t be able to visualize the results.
Research Work 3: Dynamic Pricing of Electricity: interest & limits (2022)
PAPER IN FRENCH
In this work we study the relevance on dynamic pricing in which small consumers would be exposed to variable electricity price. The price structures analysed in our document range from Time-of-Use and Peak-Time-Rebates to Increasing-Block-Tariff-Rate and Critical Peak Pricing as well as Real-Time Pricing. We study experiments that have been conducted in Spain, Ethiopia, Norvegia, France and the US, to assess the interests and limits of these pricing method. This project was conducted during the Msc. in Energy Systems Optimization of Mines ParisTech in collaboration with Ms. Simona MILADINOVA It is the result of a 4 months project as part of the energy economics course taught by J. PERCEBOIS, director of the Centre de recherche en économie et droit de l’énergie (CREDEN), and F. MIRABEL, dean of the faculty of economics of Montpellier, France.
Key take aways :
- Variable pricing can send strong price signals and incentive to load shifting
- Reflecting the real price of electricity can help investments in renwables energies, storage systems and avoid additional capacity deployment
- Real-Time-Pricing transfers all the risk to small consumers and exposes them to very high prices
- Methods such as CPP and VPP are a good compromise between consumers and electricty providers to minimize price risk and secure solvency of suppliers
Project 5: Development of a MILP to optimize the operational cost of power generation assets (2021)
if link doesn’t work, click HERE to access project
Optimal planning and operation of a power generation fleet consists in deciding on the operating schedule of the generation units to meet the forecasted load over a given time horizon, taking into account the technical constraints that govern the operation of the units while minimizing operating costs. In this project we study the development of a Mixed-Integer-Linear-Programming (MILP) model to optimize the management of 27 thermal power plants and 2 hydroelectric plants to meet a specific demand at an hourly timestep. We present the entire model as well as the code developed in python.
This project was conducted during the Msc. in Energy Systems Optimization of Mines ParisTech in collaboration with Ms. Simona MILADINOVA and under the supervising of Ms. Sophie DEMASSEY, associate professor at Mines ParisTech. This work resulted in a detailed report that is not presented in the notebook.
Project 4: Forecasting Monthly Energy Consumption in Siberia (2021)
Project Goals
Perform time series forecasting with:
Project 3: Exploratory Data Analysis of 50 hard drives available storage to prevent saturation (2021)
In the era of big data, where the benefits of data are too often emphasized, we often forget the issue of storage. With 1 million trillion Kbytes of data generated every day, we need to be able to size storage systems adapted to our needs and avoid saturation while limiting the resources used. In this project, we are exploring the availability rates of 50 different hard disks to study their evolution and establish the best models to provide information on their saturation date.
This study was conducted by Ms. Simona Miladinova and myself as part of the CIX research program of INSA Lyon in collaboration with the company INFOLOGIC, which distributes an ERP software for the food industry and provided us with their data.
It was carried out under the direction of Mr. Mehdi Kaytoue and Mr. Anes Bendimerad, both members of the R&D section of INFOLOGIC.
Project 2: Predicting the energy output of a power plant with multiple linear regressions (2020)
if link doesn’t work, click HERE to access project
Project Achievements
- Built two multiple linear regression models with both Scikit-Learn and Stats Model
- Verified linear regression assumptions (linearity, residual mean of 0, homoscedasticity, abscence of residual correlation etc.)
- Analysed multi-collinearity in features using variance inflation factor method
- Tested homoscedasticity hypothesis with Breusch-Pagan test
Project 1: Identifying a household’s energy fingerprint with KMeans clustering (2020)
if link doesn’t work, click HERE to access project
Project Achievements
- Processed 4 year historical data from an electric meter at a 1 minute timestep
- Used KMeans algorithm in order to identify electricity usage profiles of a household
- Found optimal number of centeroids with silhouette method
- Validating results with t-SNE dimensionality reduction
Research Work 2: Big Data at the service of socio-economic development in poor countries (2019)
PAPER IN FRENCH
Abstract
We often hear about Big Data and AI as technologies for maximizing corporate profit or for mass surveillance. But who would have thought that these new tools could be useful for the socio-economic development of developing countries. This is what this research paper is trying to achieve by exploring how machine learning and data science can contribute to the development of poor and developing countries or prevent humanitarian crisis situations.
This work was carried out as part of INSA Lyon’s Personal Project in Humanity (PPH).
Research Work 1: Study of a Neural Network for image classification of the MNIST database (2019)
PAPER IN FRENCH
Project Achievements
- Analysed a neural network developed in MATLAB and C++
- Performed a comparative study of the performances of both programming languages
- Used Stochastic Gradient descent and mini-batch method
- Studied optimal number of layers and node to perform classification
This work was carried out as part of INSA Lyon’s CLANU Project in collaboration with Ms. Simona MILADINOVA(Linkedin).
About Me
Seydou DIA
I am an engineer who likes to contribute to Data Science projects in relation with the energy sector and the development of Smart Grids. Mindful of the current climate and economic challenges, I firmly believe that the energy transition will be made with a smarter and above all more efficient way of consuming energy.
I hold a Meng from INSA Lyon and a Post Master Degree from Mines ParisTech. My skills range from data science to operations research as well as energy economics and financial assessment of energy projects