yiyangshen1228/Tweet-Sentiment-Analysis

Developed predictive models for tweet sentiment analysis (positive or negative). Utilized a provided dataset with labels indicating tweet sentiment and the type of English (African American English or Standard American English). Explored the performance differentiation of models across different English-speaking social groups. { "cells": [ { "cell_type": "markdown", "id": "7cbd7f7e", "metadata": {}, "source": [ "# README" ] }, { "cell_type": "markdown", "id": "9e3235a4", "metadata": {}, "source": [ "### Backgroud" ] }, { "cell_type": "markdown", "id": "0be3d057", "metadata": {}, "source": [ "This project use sklearn and other libraries to run different machine learning models to analysis the sentiment of Tweets whether belong to positive or negative. Also use semi-supervised learning model to find out whether the unlabeled data can improve the sentiment classification of Twitter or not.\n", "\n", "The code is divided into the following steps：\n", "1. Import packages and libraries.\n", "2. Read data set.\n", "3. Feature engineering and data cleaning.\n", "4. Import machine learning models and make prediction.\n", "5. Compare the results and find the best performance one for semi-supervised learning.\n", "6. Get the results from semi-supervised learning model and compare with the supervised model results.\n", "7. Use the best performance model to predict the test dataset and upload the results to Kaggle." ] }, { "cell_type": "markdown", "id": "da759e40", "metadata": {}, "source": [ "### Import packages" ] }, { "cell_type": "markdown", "id": "2c498810", "metadata": {}, "source": [ "Here are the packages that this project use:\n", "pandas,\n", "re,\n", "numpy,\n", "string,\n", "nltk,\n", "textblob,\n", "sklearn,\n", "matplotlib." ] }, { "cell_type": "markdown", "id": "3df65fc2", "metadata": {}, "source": [ "### Read data" ] }, { "cell_type": "markdown", "id": "289f021a", "metadata": {}, "source": [ "This project use two kinds of data set, one is the raw tweet data set, the other is Embedding tweet data set. All of these data set split into three data sets, one for training the ML model, the other for developing the ML model, the last is going to test the ML model." ] }, { "cell_type": "markdown", "id": "685a047f", "metadata": {}, "source": [ "### Feature engineering and data cleaning" ] }, { …