Coursework research is a research of dependence of suicidal posts in Twitter and official suicide countries rates.

It is mainly directed to examine Tweets' hashtags connected to suicidal thougths, gather data based on Tweets location or user location if the first one is not pinned. Then, creating statistics by amount of posts in various countries. The next step is to compare official data and the one that was gathered from Twitter.

FOLDERS

data

Twitter_data.db is database with tweets collected from Twitter with twitter_db.py module help. It has around 18 thousand tweets with #suicide.

Official_data.db is an official database containing information about suicides all over the world. daily_reports.json is a .json file, where there is some information about tweets amount in various countries depending on the day of a week.

Cities_data.db, consists of a country, its cities and code of a country.

tweets_amount.json contains some information about a number of tweets in specific countries.

data_preparation

This folder is concentrated mainly on gathering, grouping and selecting relevant data, that is in folder data.

Official_data.db was formed with beautifulsoup_parsing.py help, where I have used Beautiful Soup library to parse a table on a website.

ADT_Tweets.py is directed to count amount of tweets in each country. It uses both geonames library and created database of cities and its countries to ensure, that the information will be relevant. Additionally, geonames has limited amount of requests and as a consequence, I had to form Cities_data.db, which consists of a country, its cities and code of a country. It was formed with cities_db.py module. Folder has a testing_ADT.py module, which actually tests an ADT.

A file testing_api.py is testing Twitter API and abilities, that can be presented by tweepy library in working with Twitter API. Unfortunately, I cannot share a hidden module in case of private policy, that was signed when getting Twitter API keys.

It also has functions to write gathered data into .json and .csv files.

Twitter_data.db is database with tweets collected from Twitter with twitter_db.py module help. It has around 18 thousand tweets with #suicide. twitter_db.py module rewrites data from a .txt file with tweets, forming a database.

analysis

week_tweets.py module is directed to analyse tweets "behaviour". It is analysing the frequency of posting tweets at a specific day, groups tweets by countries in a specific days, finds out in what week day there were the majority of tweets posted. There is a week_tweets_test.py, which is a testing module for week_tweets.py, that uses unittest to check a program working correctly.

countries_correlation.py is a module, that counts coefficients of correlation for each country firstly depending on official data and then on Twitter-gathered one. It makes a simple counting like gets the last country in rates of suicides and compares a number of suicides of each country to the last one. Then it does the same, but with the amount of tweets. If some kind of correlation would exist, those coefficients should have been more or less similar.

spearman_correlation.py is a module to calculate the Spearman correlation coefficient. It shows if something depends on something else. It has a quite simple formula. The coefficient can vary from -1 to 1. If a coefficient is > 0.5 - the correlation is high and low otherwise.

main.py is the main module that visualizes results of a research.