Social Media AI Sentiment Analysis

This repository contains Python files for generating synthetic social media post data and analyzing the dataset.

Features

• Data Generation: Generate synthetic data to perform analysis on.
• Data Analysis:
  • Generate summary of the data.
  • Plot data in many different ways.
  • Analyze how data visualization can be misleading when focused on one subset of the data.
• Prediction using Linear Regression
  • Use sklearn's Linear Regression to predict future sentiments on historical data.

Setup Instructions

1. Install the required libraries. You can install the necessary libraries by running the following command:

pip install -r requirements.txt

2. Generate the dataset The dataset can be generated by running the generate.py file. A generated dataset "posts.db" comes included with 10000 observations.

• database_size Number of observations in the generated database.
• percent_twitternegative Percentage of posts to generate with "Negative" sentiment on "Twitter" platform.
• percent_instanegative Percentage of posts to generate with "Negative" sentiment on "Instagram" platform.
• percent_facebooknegative Percentage of posts to generate with "Negative" sentiment on "Facebook" platform.
• start_year Starting year for dataset.
• end_year Ending year for dataset.
• pos_increase How intense the number of positive posts increases per year. Higher = bigger increase. Number between 0 and 1, can be negative as well.

3. Generate data summary To gain a general overview of the data, run the summary() function in main.py. This will create a .txt file called "summary.txt" which contains general information about the data. A summary.txt file comes already included on the generated dataset.

4. Generate plots To better understand the generated data, run the plot_data() function in main.py.

5. Run predictions using Linear Regression with sklearn To see how social media sentiments might change in the future based on historical data, run the predict_future() function in main.py.