Skip to content
This repository has been archived by the owner on Oct 29, 2019. It is now read-only.

Built a twitter sentiment analyzer to determine what people were saying about companies. Then wrote a prediction app to buy or sell stocks. Ended up being a failure and wrote a paper about why it does not work. Submitted as a part of my masters

Notifications You must be signed in to change notification settings

nbroeking/Twitter-Data-Mining

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

80 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

DESCRIPTION

In this project we created a sentiment analyzer that streams twitter collecting tweets on keywords related to Apple, Google, Samsung, and Amazon. We then took these tweets ran them through a Naive Bayes Classifier. We then took the classification ran correlation analysis and determined that there is no correlation between the two data sets. In addition to the analysis we also created a real time sentiment engine. This engine creates a Rest Api that allows remote devices to get real time sentiment analysis. To compliment the api there is a android application that will reach out to the api and process the information and display it to the user.

Report

The report, which includes key results and information is included in Report.pdf. All resources used to create the report can be located in the sub director Report.

Analysis

The first stage of our process includes all code related to the correlation. This code can be found in the Analysis sub folder. It consists of several sections.

Streaming.py

This file has all the code that pulled tweets and stored them in our database.

Classifier.py

This file pulls the tweets from the database. Runs the pre-processor and classifier and then exports the data to a csv file.

Stock_tweets.r

This file contains all the code used to determine no correlation between the data sets.

training.csv

This file is read in by the classifier and contains all the training tweets we used to train the Naive Bayes Classifier.

stop.txt

This is a list of stop words that are determined to have no sentiment and so we can remove them from our feature vector.

Realtime

This sub directory contains all the code related to the Application. TwitterClassifier contains all the Android code and restApi contains all the server code.

About

Built a twitter sentiment analyzer to determine what people were saying about companies. Then wrote a prediction app to buy or sell stocks. Ended up being a failure and wrote a paper about why it does not work. Submitted as a part of my masters

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published