Sergio Vechi's Projects
What I learned when trying to transcribe an audio file with Python
Testing
Explotarory Data Analysis on the cancer hospital records database, a dataset formed from data generated by several hospital institutions located in the State of São Paulo (Brazil).
The probabilities calculated by the model BG-NBD are used to define the target variable to predict customer churn.
Implementation of COIN from the paper "COIN: COmpression with Implicit Neural representations" .
Predicting crime rate using DecisionTreeRegressor(), analyzing the importance of specific features, and reducing the complexity of the model.
Exploratory data analysis, data cleaning, and feature engineering were performed, then regression models using linear regression, decision tree, random forest and xgboost were built and evaluated. The customer segmentation was addressed using k-means clustering.
Exploratory data analysis to probe Kleinanzeigen used car ads. How mileage influences the price?
Predicting employee retention using LogisticRegression() classifier, exploring metrics, building ROC curves.
Exploratory data analysis to investigate trends in the data set of the Hacker News site. Which posts (ask or show) receive more comments on average? Is there a best time to post?
Testing options
Selecting and training a classification model to predict the risk for heart disease, putting it to a web service using Docker and BentoML, and deploying to a cloud platform.
Mapping apps for profitability at the App Store and Play Store. How data analysis of app profiles can help in improving the profitability of the business?
Created for testing
Several baseline models were built to predict the survival rate of patients after one year of treatment. The best ranked by the F1 score were selected and improved since this was the evaluation criteria for this task.
Mini Competition in DSR Batch 25: prediction of Rossman sales
Predicting gender from eye fundus images using deep learning.
Imagine you are all about protecting the environment, trying to address some problems of fast fashion, and want to wear sustainable clothing only. You care a lot about fashion, too! But it's hard to find sustainable pieces that match the ideas you have for your look. Imagine you could take a picture of a trendy person on the street and get recommendations on what they are wearing, but from a second-hand market. This is what our app deep fashion does.