📊 Data Science Portfolio

A curated collection of Data Science projects covering Machine Learning, Natural Language Processing, Social Network Analysis, and Data Visualization.

---

📁 Repository Structure

DataScience/
│
├── MachineLearning/
│   ├── Clustering/             # K-Means consumer segmentation (R)
│   ├── PLN/                    # NLP: Collocations, BOW, TF-IDF, Word2Vec, GloVe, CBOW (Python)
│   ├── SVM/                    # Support Vector Machines: multiclass & e-commerce (R & Python)
│   └── Sistemas_Recomendacao/  # Recommendation systems (Python)
│
├── SocialNetworkAnalytics/
│   └── Twitter/                # Twitter social network analysis (R)
│
└── VisualizacaoDados/
    ├── DashboardDesign/        # Interactive dashboard design (R Shiny)
    ├── Metodos_Visualizacao/   # Visualization methods and techniques (R)
    └── Visualizacao_D3.js/     # Data visualization with D3.js (JavaScript)

---

🤖 Machine Learning

📌 Clustering — Consumer Segmentation with K-Means

Language: R | Algorithm: K-Means

Identify consumer segments for personalized marketing campaigns using clustering techniques.

Pipeline:

• Exploratory Data Analysis (missing values, visual analysis)
• Preprocessing: variable standardization and Hopkins statistic for cluster tendency
• Modeling: NbClust for optimal cluster number selection + K-Means cluster analysis

---

📌 NLP (Natural Language Processing)

Language: Python | Libraries: NLTK, PyTorch, SciPy

A series of NLP projects covering modern text analysis techniques:

Project	Description	Technique
Hotel Reviews Collocations	Find relevant bigrams/trigrams explaining hotel ratings	PMI, t-test, Chi-square
Language Classifier	Classify sentence language from trained model	Bag of Words + PyTorch
Keyword Identifier	Identify most relevant words in a book	TF-IDF
Sentence Predictor	AI for legal text sentence prediction	CBOW
Word Similarity Search	Find similar words using word vectors	GloVe
Word Context Predictor	Word prediction with PCA visualization	Word2Vec

---

📌 SVM — Support Vector Machines

Language: R & Python | Libraries: e1071, scikit-learn

Project 1 — Multiclass Credit Card Spending Classification (R)

• Missing value imputation with PMM (mice)
• Class balancing with SMOTE (DMwR)
• Comparison of Radial, Linear, and Polynomial kernels with GridSearch

Project 2 — E-Commerce Purchase Intent Prediction (Python)

• Label encoding and SMOTE oversampling
• Training time optimization: from 4,003s → 7s
• Hyperparameter tuning with GridSearch (RBF and Polynomial kernels)

---

📌 Recommendation Systems

Language: Python | Notebook: Jupyter

Building recommendation engines to suggest items based on user behavior and preferences.

---

🌐 Social Network Analytics

📌 Twitter Network Analysis

Language: R

Social network analysis of Twitter data to explore connections, influence, and community structure.

---

📈 Data Visualization

📌 Dashboard Design

Language: R (Shiny)

Interactive dashboards built with R Shiny for exploratory data analysis and reporting.

📌 Visualization Methods

Language: R (ggplot2, plotly)

A collection of visualization techniques including statistical plots, heatmaps, and interactive charts.

📌 D3.js Visualizations

Language: JavaScript (D3.js)

Dynamic and interactive data visualizations built with D3.js, including bar charts and force-directed graphs.

---

🛠️ Tech Stack

Area	Tools
Languages	Python, R, JavaScript
ML & DL	scikit-learn, PyTorch, e1071, DMwR
NLP	NLTK, SciPy
Visualization	ggplot2, plotly, D3.js, R Shiny
Notebooks	Jupyter Notebook, R Markdown
Data Handling	pandas, dplyr, mice

---

🚀 Getting Started

Prerequisites

• Python 3.7+ or R 4.0+
• Jupyter Notebook

Running a Project

# Clone the repository
git clone https://github.com/Thaleslsilva/DataScience.git
cd DataScience

# Navigate to a project folder, e.g.:
cd MachineLearning/SVM

# Open the notebook
jupyter notebook

---

📬 Contact

Thales de Lima Silva

---

Suggestions and contributions are welcome! Feel free to open an issue or submit a pull request.