Small Data Science Projects (2014-2016)
-
Undergrad_FinalProject_2014: Final Project required for BS in Statistics
Supervisor: Fabio Demarqui
Project: Generalized Linear Model to evaluate the students performance in Calculus class
Summary: In this project we compare the performance of students from different departaments in the Universidade Federal de Minas Gerais using a GLM model. Note: Report is in Portuguese -
GameOfThrones_2016: Graphics and other descriptive statistics about a Game of Thrones dataset
Data Source: https://kaggle.com/mylesoneill/game-of-thrones
Note: Report is in Portuguese -
Undergrad_ResearchProject_2014: Item Theory Response(ITR) analysis to measure the political knowledge of university students in Brazil
Data: We collected the dataset for a course project
Summary: Descriptive Analysis and ITR models to compare the political knowledge of university students.
Note: Report is in Portuguese -
ML_FinalAssignment_2015: Machine Learning Course Project - 2015 - UFMG
Group: Raquel Aoki and Fabrício Rodrigues Inácio
Project: Classification model of supermarket products (Source: https://www.kaggle.com/c/otto-group-product-classification-challenge)
Summary: The goal of this project was predict 9 labels from the dataset with 93 features. The F1 score was used to compare the approaches due the imbalance between classes and 2 dimensionality reductions methods were adopted: PCA and RBM. For the classification part we used SVM and we compare the results on the traning and testing set between SVM+93 original features with SVM+PCA features and SVM+RBM features. We also tested different dimensions in the features set from the PCA and RBM. The conclusion was that the SVM using the original features have the best results on the training set, but also suffers with overfitting, as we can see on the poor performance on the testing set. The SVM+63 features from the PCA has the best overall results. Note: Report is in Portuguese -
Stats4Good: Workshop Stats4Good presentation - 2016 - UFMG
Summary: Short description about me and my Undergrad_ResearchProject_2014.
Code: Plots made speficit for this presentation
Note: Presentation is in Portuguese