lykability 1.0.0

Using empythy to score likability based on sentiment analysis of recent tweets about a given person

Using empythy to score likability based on sentiment analysis of recent tweets about a given person

Purpose

To piggyback off of the empythy natural languare classifier package to analyze average sentiment of tweets related to a particular person to calculate a ‘likability score’ for that person. Useful in tracking sentiment changes across a certain period of time, i.e. the likability score of a celebrity before and after a concert.

Instructions

• Open terminal. Make sure you have python3 and pip downloaded.
• pip install likability
• Create a csv file with the names of the people you’d like to analyze for likability. Name this file name.csv in the current directory.
• Determine how many recent tweets you’d like to query for each person. This will be used in the script below as num_tweets.
• Make sure you have Twitter API keys and access tokens. If you do not, go to [Twitter Apps](https://apps.twitter.com/), create an app, and find the required keys and tokens under Applications Settings -> Consumer Key (API Key) -> manage keys and access tokens.
• Run Python 3 by typing python into the terminal.
• Enter script below to run the LikabilityAnalyzer module.
• from likability import LikabilityAnalyzer
• filepath = 'name.csv'
• num_tweets = 100
• sentimentScore = LikabilityAnalyzer.analyzer(filepath,num_tweets)
• When prompted, enter in your Twitter API keys. This will allow likability to access the Twitter API to query the tweets needed to complete the sentiment analysis.
• Wait for script to run to completion. Please note, due to Twitter API Rate Limiting, querying more than 15 names will lead to longer wait times. Please allow 1 minute per name for lists greater than 15 names.
• Upon completion, open the newly created Sentiment.csv in the current directory to access the likability scores for each person.

Possible Usage

• Score top fantasy football players to see what the Twittersphere thinks about each player pre-draft
• Instead of names of people, use product names to track customer sentiment in real-time
• Solve the question: who is more likable, Justin Timberlake or Jimmy Fallon