/ENSAE_projects

School Statistics and Data Science Projects

Primary LanguageJupyter NotebookGNU General Public License v3.0GPL-3.0

University Projects

Data Science

  • Supervised_Classification : Machine Learning Project of Predicting the contraception choice of Indonesian Women.
    Language : Python
    Date : November 2015
    Methodology : Comparison of Machine Learning Supervised_Classification algorithms.

  • ParallelizedPCA : Parallel GPU Implementation of Iterative PCA Algorithm
    Language: pyCUDA Date: january 2017
    Team: Gauthier Schweitzer
    Methodology: Gram-Schmidt orthogonalization algorithm proposed by Andrecut (2008)

  • TopicModelingLDA : Discussion and Implementation of Topic modeling in text analysis by Latent Dirichlet Allocation
    Language: Python
    Date: January 2017
    Team: Chloé Baraille, Othmane Sebbouh
    Methodology: collapsed variational bayesian inference algorithm for Latent Dirichlet Allocation (Teh, Newman, Welling, 2006)

  • NetworkLasso : Predicting prices in the Housing Market of Sacramento
    Language : Python
    Date : June 2016
    Team : Victor Cluzel, Noémie Haouzi
    Methodology : Network Lasso Optimization

Econometrics

  • SemiParametricTilting: Resolving Semi Parametric Data Combination Issues.
    Software: Python
    Date: January 2017
    Team: Leïla Fardeau, Loïc Macherel
    Methodology: Auxiliary-to-Study Tilting (Graham et al. 2016)

  • Selection_Bias&Panel_Analysis : Discussion on the evaluation of a causal relationship between Democracy and Income.
    Software : STATA
    Date : March 2016
    Team : Noémie Haouzi
    Methodology : Cross-Country Panel Analysis, Selection Model, Instrument Variables.

  • SyntheticControlMethod: Evaluation of the impact of a cultural policy (*Lille European Capital of Culture 2004) on regional economic indicator (GDP and Unemployment rate).
    Software : R
    Date : January-March 2016
    Team : Louise Aimène and Marc Ginestet
    Methodology : Synthetic Control Method of Abadie, Diamond and Hainmueller.