Hello! Welcome to my data science projects portfolio.
My name is Jaime. I am a professional in Data Science and here you will find a list of projects that I have worked related to this field.
I consider myself as a continuous learner and also, I like to share my knowledge to those who are just starting on this broad field. Data science has a vast list of applications to any kind of industry and keeps growing every day. If you have data in your hands, you can start applying data science with one simple goal: extract meaningful insights in your data.
1. Business Time Series Forecasting
In this document, you will find a business forecasting application to predict sales revenue. This is achieved by applying machine learning algorithms such as Neural Networks & XGBoost, but I also explore tradicional (sequential) models such as ARIMA and its variants. I use tools such as "tidymodels" and "modeltime" from R Programming, and the integration of several forecasting packages in R. For deployment, I have used Markdown, a powerful tool to share documents via HTML.
A very useful application in the day-to-day business is to assess our employee turnover. The goal of this project is to apply supervised machine learning algorithms: Logistic Regression & Decision Trees, to predict the likelihood of attrition from a specific HR data. Here, I am focused in using Python and its vast packages for data science such as Pandas, NumPy and Scikit-learn. In addition, I have used "Reticulate" package, which is capable to execute code from Python and R in Rstudio ide. A Jupyter Notebook is also available
3. Default loan prediction (Fintech)
This application is intended to show how machine learning can be used in several industries to solve real problems and take decisions. Digital lending industries and FINTECHS are some examples, and the goal of this application is put an algorithm in production called Random Forest that can help to estimate what is the probability that a borrower is likely to pay a loan. In other words, how likely is that borrowers fail to pay back a debt according to an initial arrangement.
This application was designed using the open source software “R” and Shiny dashboard, which is one package that can be attached to it. Other packages such as H2O and Leaflet were used
4. Modelo de predicción de fuga de clientes (spanish)
Un banco del sector desea crear un modelo predictivo que permita estimar la probabilidad de fuga de cada cliente para la cartera de créditos proporcionada. El objetivo es crear un modelo de aprendizaje automático que permita clasificar dicha fuga de clientes para cada registro. Un modelo de predicción de fuga permitirá al negocio, el poder monetizar la predicción obtenida y estimar una posible pérdida monetaria en la cartera de créditos.
Here you will find a list of the tools that I use to do data science. Right now, there are several options on the market but the ones that stand out are: Python, R and Julia. These are open-source tools, which means that you don't pay a single dime to obtain them.
- R Programming: Official site of R
- Python Programming: Official site of Python
- Tidymodels: Build machine learning models
- Modeltime: Another tool to build ML models
- Pandas: Wonderful package to analyze data in Python
- scikit-learn: Set of advanced packages to apply ML/Deep Learning
- NumPy: Package for linear algebra
- H2O.ai: Build automated machine learning applications!
- Shiny: Great interface for dashboards and WebApps in R
- Markdown: Publish documents in R/Python and build presentations!
- Jupyter Notebook: Most recognized interface to execute Python code
- RStudio: R interface to execute R code with easy!
- Anaconda: Popular Python/R platform for deployment
Follow me on @Linkedin Contact me at gmail - [email protected]