L-N-E0/USAID-Kenya-Sentiment-Analysis

NLP project analyzing Kenyan public sentiment on USAID funding cuts using Reddit, X, and news APIs. # USAID Online Discourse Analysis ## Project Overview This project analyzes online discussions related to **USAID in Kenya** after the funding cuts, using data collected from **Reddit** and **News API** sources. The goal was to identify **key topics, trends, and sentiment patterns** in public discourse and provide insights to inform policy, outreach, and communication strategies. The analysis was conducted as part of a collaborative group project. While the dataset selection was decided collectively, **data collection, merging, cleaning, analysis, and visualization** were done individually. This ensured methodological diversity while working with a shared dataset. --- ## Steps Followed 1. **Data Collection** – Gathered USAID-related content from multiple online sources. 2. **Data Cleaning & Preprocessing** – Removed noise, normalized formats, and prepared text for analysis. 3. **Exploratory Data Analysis (EDA)** – Analyzed distributions, trends, and word frequencies. 4. **Sentiment Analysis** – Measured polarity trends over time. 5. **Topic Modeling** – Extracted and interpreted key themes. 6. **Visualization & Communication** – Presented findings through clear, accessible visuals. --- ## Data Sources - **Reddit API** – Discussion threads and comments. - **News API** – Articles and headlines referencing USAID. --- ## Tools & Technologies - **Python** – Core programming language for analysis. - **Pandas / NumPy** – Data manipulation and numerical computation. - **NLTK / spaCy** – Natural Language Processing (tokenization, lemmatization, stopword removal). - **VADER** – Lexicon-based sentiment analysis. - **LDA / BERTopic** – Topic modeling to uncover latent themes. - **Scikit-learn** – Machine learning workflows. - **Matplotlib / Seaborn / Plotly** – Visualization tools. - **Tableau / Power BI** – Interactive dashboards. - **Jupyter Notebook** – Experimentation, prototyping, and documentation. --- ## Findings & Insights - **Dominant Themes** – Con …