My Projects by Roberto Calzadilla

NLP PROJECT - SENTIMENT_CLASSIFICATION

Project overview

• Builded a ML model to classify mental health status from text data using NLTK and XGBoost.
• Preprocessed the text, did feature engineering and apply TF-IDF as bag of words
• After tunning I saved and tested the model with a 82% accuracy

You can access to it HERE

ML PROJECT - RECOMMENDATION SYSTEM

Project overview

• Developed a movie recommendation system using matrix factorization techniques.
• Compared two methods: Regularized Least Squares implemented from scratch versus SVD using the Surprise library.
• The scratch-built model slightly outperformed the library-based model, but with higher complexity and performance.

You can access to it HERE

RAG PROJECT - TRAFFIC REGULATIONS CHATBOT

Project overview

• Developed a chatbot application to help users to clarify questions about Mexico City’s traffic regulations
• Leveraged LangChain, AWS Bedrock, and OpenSearch for the backend infrastructure.
• Implemented the Retrieval-Augmented Generation methodology, using Titan for embeddings and Claude Haiku as the Language Model.

You can access to it HERE

DATA ENGINEERING PROJECT - HOUSE PRICES

Project overview

• Developed a complete ETL process using different technologies (Pyspark, beatifulsoup, AWS, sklearn)
• Scraped information from a real state webpage, the crude data is stored in S3
• Cleaned the data with AWS Glue and pyspark, then loaded it to Athena SQL
• Finally did EDA and model a linear regression to calculate the missing data about total area

You can access to it HERE

DATA ANALYSIS PROJECT - SALES MANAGEMENT

Project overview

• This projec explores different business metrics and how they can be displayed graphically to follow them
• I explore the relation between sales, budget, costumers and products
• We can obtain conclusions that will help us to make better marketing campaigns and improve profits
• The tools used are SQL server and Power BI

You can access to it HERE

EXCEL DATA ANALYSIS PROJECT

Project overview

• An electronics store needs to increase its profits
• A KPI is found that will allow to increase profits at the lowest cost.
• Sales were evaluated by product, city, time and over time.
• This analysis was done on a database with 180,000 values using Excel only.

You can access to it HERE

PYTHON - BANK LOANS ANALYSIS

Project overview

• We need to know which factors influence to give a loan
• We also need to look for any opportunity to increase the bank’s assets.
• It was found that apart from income, family size is an important factor to give loans.
• The libraries used were pandas, numpy, matplotlib, seaborn and scipy.

You can access to it HERE