MENU LIST

Miyah Segura’s Data Science Portfolio

Contact Information

About Me I am a data-driven professional with a Master’s in Data Science and 5+ years of experience in data analysis, system integration, and workflow automation. My expertise spans across SQL, Python, Power BI, and cloud-based platforms, allowing me to develop scalable solutions that drive business insights and efficiency. This portfolio showcases my data science projects, highlighting my technical skills and problem-solving approach.

Projects

1. Angel Number Identifier Web App

Key Achievements: Gained significant engagement from the spiritual community and successfully deployed the app using Render.

Tech Stack: Python (Flask), HTML/CSS, OpenAI API, C#, SQL

Description: Built a web application that interprets angel numbers, providing users with spiritual insights based on numerical patterns. Implemented API calls to OpenAI for dynamic interpretations and integrated SQL databases for efficient data handling.

Try the Site: angelnumberidentifier.onrender.com

Github: Angel Number. PY File

2. Face Detection and Recognition Using Deep Learning

Overview:

This project applies Machine Learning and Computer Vision techniques to detect and recognize faces of baseball players from the 2024 Philadelphia Phillies. The dataset includes images extracted using the MTCNN face detection algorithm and processed using TensorFlow and Scikit-Learn.

Key Highlights:

✔ Face Detection: Used MTCNN to extract facial regions from images.
✔ Feature Extraction: Leveraged ResNet50 to obtain feature embeddings.
✔ Classification Model: Trained an SVM classifier to recognize individual players.
✔ Model Evaluation: Assessed classification performance using Precision, Recall, and F1-score.
✔ Deep Learning Integration: Implemented TensorFlow for image preprocessing and model training.

Technologies Used:

Data Processing: Image Preprocessing, Data Augmentation

Programming Languages: Python

Libraries & Frameworks: TensorFlow, OpenCV, Scikit-Learn, MTCNN

Machine Learning Models: SVM, ResNet50 Feature Extraction

Github: Face Detection Jupyter File

3. Unigram Language Model for Text Analysis

Overview:

This project focuses on building a unigram language model to analyze and predict word probabilities within a text corpus. The model was trained on Bram Stoker’s novel “Dracula”, using text preprocessing, tokenization, and frequency-based probability calculations.

Key Highlights:

✔ Text Preprocessing: Cleaned and tokenized text data by removing punctuation, stop words, and converting to lowercase.
✔ Unigram Model: Implemented a Maximum Likelihood Estimation (MLE) unigram model to predict word probabilities.
✔ Statistical Analysis: Calculated word frequency distributions and probability estimates.
✔ NLP Libraries: Utilized NLTK for text processing and modeling.

Technologies Used:

Data Processing: Tokenization, Stopword Removal, Frequency Analysis

Programming Languages: Python

Libraries & Frameworks: NLTK, NumPy, Pandas

Machine Learning Models: Unigram Language Model (MLE)

Github: Unigram Jupyter Notebook