Diego Villacreses
About
Detail
Científico de datos / Economista
Provincia de Pichincha, Ecuador
Contact Diego regarding:
work
Full-time jobs
Starting at
USD1.7K/month
Timeline
work
Job
school
Education
folder
Project
Résumé
Jobs
verified_user
40% verified
Consultor en analítica e inteligencia artificialKin AnalyticsMay 2022 - Jul 2022 (3 months)- IStatistical analyst verified_user Verified experienceInstituto Nacional de Estadística y Censos (INEC)Jun 2017 - May 2022 (5 years)- Investigador. Temas principales: Machine learning para selección de variables y predicción. Estimación de áreas pequeñas. Modelos de variables latentes. Inferencia y análisis en muestras complejas. Investigador. Temas principales: Machine Learning para selección de variables y predicción. Estimación de Áreas Pequeñas. Modelos de Variables Latentes. Inferencia y Análisis en Muestras Complejas.
- Especialista en macroeconomíaBanco Central del EcuadorJun 2016 - May 2017 (1 year)
- FFreelance verified_user Verified experienceFreelancerJan 2016 - Current (9 years 7 months)
- ESenior economistECONOMICA CIC - Centro de Investigación en economía, finanzas y política públicaJan 2013 - Jul 2015 (2 years 7 months)
Education
verified_user
0% verified
- Micro masters program in statistics and data scienceMITx on edxSep 2022 - Current (2 years 11 months)
- Neural Networks and Deep LearningDeeplearning.aiAug 2022
- Deep learning.Ai TensorFlow developer public Remote experienceDeeplearning.aiAug 2022 - Sep 2022 (2 months)
- IMacroeconometric Forecasting public Remote experienceInternational Monetary FundJan 2017 - Apr 2017 (4 months)
- Bachelor of science - BS, economía y econometríaPontificia Universidad Católica del EcuadorAug 2009 - Jun 2015 (5 years 11 months)
Projects
verified_user
0% verified
- FFull Dollarization versus Monetary Union: A look at the Ecuadorian CaseFreelancerSep 2022 - Mar 2023 (7 months)- Co-author - The finding of this study clearly shows that the Ecuadorian FD did change its monetary relation with the US economy. In the one hand, post-dollarization, not only did US and Ecuadorian inflation rates start to synchronize, but they also did converge. - We present preliminary evidence about the fact that inflation synchronization indeed happened ex-post the Ecuadorian FD (or IMU) adoption, and consequently, monetary policy changes of US do affect Ecuadorian macro-variables patterns. - However, OCA Theory is indeed a very useful framework to study FD regimes. And we found evidence to use USA Intereset Rate in Ecuadorian macroeconomic models.
- PCredit card default prediction using various approaches to assess class imbalanceProfesional independienteMay 2022 - Jul 2022 (3 months)- This is a notebook detailing the implementation on Python of six models to maximize a Bank Profit function under heavy class imbalance and compare it to other standard gain and loss functions. - We used information of 30,000 Taiwan's customers produced on October 2005, detailed description of the information we used can be found on the for Machine Learning Repository of University of California, Irvine. - We use as a Bank Profit function: , where is a parameter that allow us to impose a relative value of a default client against a non-default. Since we doesn't know $\alpha$ we would train our models with.
- FInference for Subsamples in Complex Surveys using RFreelancerFeb 2021 - Mar 2021 (2 months)- This R code computes exact Confidence Intervals for any Complex Survey under any subpopulation regarding the presence of missing data. This script replicates subpop command from Stata, based on the research presented in [1]. - Many practitioners usually oversee this issue, forgetting that the computation of adequate Degrees of Freedom (df) for Confidence Interval's t-distribution must follow special rules. - As mentioned in [1] this problem is still a matter of research. As far as I studied the literature, there aren't following investigations about this topic. - As far as I know, there isn't an R implementation about this. So, I hope someone finds this useful. - [1] West, Brady & Berglund, Patricia & Heeringa, Steven. (2008). A Closer Ex
- FA full walkthrough (I hope) of XGBoost in RFreelancerMay 2020 - Jul 2020 (3 months)- As in my recent experience I didn't found a full tutorial, walkthrough or example of how to perform a step by step "personalized" XGBoost in R, I decided to upload this. Hope the code will be useful to someone. - You could use any loss-function (evaluation metric, gain function...) to perform a fully parallelized Hyperparameter Tuning and then use XGBoost with those hyperparameters for whatever you want. - Computation of an "interpretable model" from our XGBoost (thanks to AppliedDataSciencePartners/xgboostExplainer)
Publications
verified_user
0% verified
- hhttps://papers.ssrn.com/sol3/papers.cfm?abstract_id=4365154