clydeochieng/Sentiment-Analysis-on-Kenya-Power-Customer-Experience

# Sentiment Analysis for Kenya Power and Lighting Company (KPLC) ## Business Understanding ### Overview Kenya Power and Lighting Company (KPLC) is a major utility provider that receives a high volume of customer feedback through social media platforms, particularly Twitter. Understanding customer sentiment is crucial for KPLC to improve its services, automate responses, and enhance customer satisfaction. The project aims to develop a sophisticated chatbot that can classify various types of tweets and generate appropriate automated responses. ### Problem Statement KPLC faces the challenge of efficiently processing and categorizing customer feedback from social media, especially Twitter, where customers express their sentiments about KPLC's services. By accurately classifying tweets into sentiment categories, KPLC can identify common complaints, pinpoint service issues, and enhance customer feedback mechanisms. This will enable KPLC to improve service quality, response times, and overall customer experience. ### Objectives 1. **Gauge Overall Customer Sentiment:** Understand the general sentiment towards KPLC's services to identify areas for improvement. 2. **Identify Specific Issues:** Detect and categorize specific issues mentioned in tweets, such as power outages, billing problems, and token issues. 3. **Create a Responsive Chatbot:** Develop a chatbot that can provide appropriate responses to customer inquiries, improving customer service efficiency and response times. ### Challenges 1. **Data Collection and Preprocessing:** Gathering relevant tweets mentioning KPLC, cleaning and preprocessing the data, and handling noise like unrelated tweets and abbreviations. 2. **Sentiment Analysis Accuracy:** Dealing with informal language, sarcasm, mixed sentiments, and local dialects often used on social media platforms. 3. **Identifying Specific Issues:** Extracting and categorizing specific issues mentioned in tweets can be complex due to diverse customer descriptions …