I predict Nigeria’s presidential election outcome using Twitter sentiment analysis. Data shows 42.6% positive, 13.2% negative, and 44.2% neutral sentiment. Labor party leads positive sentiment. Sentiment analysis helps understand people’s emotions and opinions. I used NLP and Vader for efficient processing. Learned to build an app, deploy on Streamlit, and work with machine learning models. Check out my project! 📊🇳🇬 #NigeriaElectionPrediction #DataAnalysis 🗳️
Overview 📊
The presentation took place on February 25th, 2023, during Nigeria’s presidential election. The aim of the project was to predict the election outcome using Twitter data to analyze people’s sentiments based on their preferred candidates. Victorio de Anderson, the presenter, is a data analyst with a background in customer service and is transitioning to becoming a machine learning engineer. The sentiment analysis involved evaluating the negative, positive, and neutral opinions or emotions of Nigerians towards the election using data gathered from Twitter.
Data Collection and Analysis 📈
The data was collected from Twitter using Essence script, which allows for the extraction of tweets based on specific criteria such as hashtags and user accounts. For this project, a total of 20,000 tweets were gathered, with 5,000 each for four different parties. The analysis revealed that 44.2% of people had a neutral stance, 42.6% had a positive sentiment, and 13.2% had a negative opinion regarding the election. The Labor Party had the highest sentiment score among the positive sentiments.
Understanding Sentiment Analysis 🧠
Sentiment analysis is a technique used to analyze the tone and emotion of text, classifying them as positive, negative, or neutral opinions. Positive opinions in text may include favorable sentiments, while negative opinions express pessimistic sentiment. The visualization presented depicted the total sentiment score by party and the total sentiment counts.
Natural Language Processing (NLP) and Sentiment Analysis 🌐
NLP is a necessary tool for sentiment analysis and machine learning. Valence (Veda) and sentiment original were used to simplify the NLP data processes for this sentiment analysis. Veda uses a pre-built lexicon that contains thousands of words and phrases to classify text into positive, negative, and neutral opinions without the need for pre-label data sets or extensive pre-processing.
Learning and Development 💻
During the presentation, the presenter shared the skills and tools learned so far, such as building a streamlined app, deploying on Streamlit and Iroko, and working with Veda machine learning model. This includes avoiding the limitations of the Twitter API, defining a project, and understanding NLP data pre-processing steps.
Deployment and Future Work 🚀
The presenter showcased their project on Streamlit, detailing the challenges faced when deploying it on different platforms. Links to the project, GitHub, and LinkedIn were also shared.
In conclusion, the KaggleX Final Project Presentation with Victory Odianosen provided valuable insights into sentiment analysis using Twitter data during Nigeria’s presidential election. The use of NLP tools, sentiment analysis, and data visualization were key components in understanding the sentiments of the people. The presenter also highlighted ongoing learning and the importance of creating reusable machine learning models.
