dfroca-portfolio

The contents of my portfolio are the following:

1 - Football Analytics

Personal project where data in real-time is extracted from an API to create a data warehouse of football scores in the European Leagues and Competitions. After this, some basic Machine Learning and Deep Learning models are created for predictive analytics purposes.

Languages and tools:

• SQL
• Pendaho Data Integration
• Power BI
• Python

2 - Data Visualization

2.1 - Youtube trending video analysis

Power BI dashboard showing youtube geographical and video category insights from a public dataset.

Languages and tools:

• Python (Jupyter Notebooks)
• Power BI

2.2 - Energetic Soda

Sponsorship company that wants to evaluate the best team to invest in based on previous team performance and expected revenews.

Languages and tools:

• Power BI

3 - Machine Learning

3.1 - Advanced Machine Learning

Coding of regression methods from scratch (Vanilla, Lasso, Ridge and M (Robust)). Coding of different clustering techniques such as K-means, Hierarchical, DBScan and HBDScan. Coding of SVM, logistic regression.

Languages and tools:

• Python (Jupyter Notebooks)

3.2 - Advanced Machine Learning

Creation of a decision tree from scratch for classification and regression using NumPy and Pandas libraries.

Languages and tools:

• Python

3.3 - Machine Learning Exercises

Banking and retail exercises to support decision making by creating ML Classification Models (Random Forest and Boost)

Languages and tools:

• Python (Jupyter Notebooks)

3.4 - Natural Language Processing

Creating a Skipgram model from scratch using NumPy and creating an Aspect-Based Sentiment Analysis using Hugging Face Models

Languages and tools:

• Python

4 - Deep Learning

4.1 - Computer Vision - Detectron V2

Detect and classify workers on construction-site images to detect potential risks. The Image detection was performed using Detectron V2

Languages and tools:

• Python (Jupyter Notebooks)

4.2 - U Net for Image Segmentation

Creation of a UNet Neural Network Architecture to segment images and classify the objects that were detected.

Languages and tools:

• Python (Jupyter Notebooks)

4.3 - Graph NN

Graph NN Architecture to predict protein characteristics (PPI Dataset)

Languages and tools:

• Python

4.4 - Generative NN

GAN, Auto-encoders and Normalizing flows Exercise

Languages and tools:

• Python (Jupyter Notebooks)

4.5 - NN Tuning

Neural Network tuning exercise

Languages and tools:

• Python (Jupyter Notebooks)

4.6 - Basic NN for Regression and Classification

Regression and Classification NN and hyperparameter tuning

Languages and tools:

• Python (Jupyter Notebooks)

4.7 - Knowledge Mining

Document extraction, named entity recognition and to identify corrupt behaviours in Colombian Public Sector

Languages and tools:

• Python (Jupyter Notebooks)
• Microsoft Azure ML Services

5 - Other

5.1 - Spark

Excercises that aim to understand the overall map-reduce paradigm and include

Languages and tools:

• Python (Jupyter Notebooks)

5.2 - Life-Time Value Analysis

Excersise for marketing, use a Pareto-NBD Model to understand the behaviour of the clients and prioritize campaigns

Languages and tools:

• R

5.3 - Time Series Analysis

Multi-variate time series analysis focused on M5 Competition

Languages and tools:

• R

5.4 - Trainings

Business Intelligence trainings, delivered in a previous work-experience.

Languages and tools:

• Power BI