Step by Step Sentiment analysis on Twitter data using R with Airtel Tweets: Part – II

In the previous post we saw what is sentiment analysis and what are the steps involved in it. In this post we will go through step by step instruction on doing Sentiment Analysis on the Micro blogging site “Twitter”. We will have specific objective to do so. I came across an interesting post by Chetan S on the DTH operators involvement in using Social Media for providing customer support. It triggered me the idea for this post.

Goal: To do sentiment analysis on Airtel Customer support via Twitter in India.

In this Post: We will retrieve the Tweets, look at how to access the Twitter API and make best use of the TwitteR R package and write these tweets to a file.

Important Note:

1. when you would like to use the searchTwitter, go to dev.twitter.com and your application go to the “Settings” tab and select “Read, Write and Access direct messages”. Make sure to click on the save button after doing this.”

Refer to this link http://stackoverflow.com/questions/15713073/twitter-help-unable-to-authorize-even-with-registering

2. When you are trying to search using searchTwitter after the above step if you get ssl problem make sure you have enable rCurl and do the steps outline here: http://stackoverflow.com/questions/15347233/ssl-certificate-failed-for-twitter-in-r.

options(RCurlOptions = list(cainfo = system.file(“CurlSSL”, “cacert.pem”, package = “RCurl”))) also make sure you have loaded the Necessary Packages like ROAuth,

Step 1: Make sure you have done the OAuth authentication with Twitter using the Previous post and the steps outlined above, you can also check the library loaded with sessionInfo(). Step 2: Make sure you load the tweets from the Twitter from the Twitter Handle accordingly > airtel.tweets=searchTwitter(“@airtel_presence”,n=1500) Now we have loaded the 1499 tweets which was responded by the Twitter API in to airtel.tweets. Now what we will do is to save these to a file for future processing. Step 3: Before we write these tweets to a file, for better understanding we will try to look at some of the tweets and data collected so far. head(airtel.tweets) provides the top 6 tweets. Further to our analysis, we try to get the length of the tweets, what kind of class it is and how can we access the tweets. Look at the below given screenshot. Step 4: We will look at some examples of How to access the twitter data in a better fashion with respect to the Twitter API using TwitteR library by accessing one tweets from the 1499 available. In this above given example we have selected the 3rd item from the list and we have tried to get till the user information, how many friends he has and how many followers he has, etc., These are the things which are vital to understand as these factors can become viral and impact the image of a particular brand. Now will go to the next step of identifying the steps to store these tweets for further analysis. Step 5: We will store these tweets we collected in airtel.tweets to a file for future analysis and reference. We are going to convert the list of tweets to separate data using apply functions and write to a file. We are going to use the library plyr for the same. Plyr allows the user to split a data set apart into smaller subsets, apply methods to the subsets, and combine the results. Please click here for detailed introduction on plyr. So we are converting the list to data frame for preparing it to be written to a file. Now the tweets and all the necessary information is available in the tweets.df data frame. You can look at the below screenshots for its summary. Step 6: Setup the Working directory and write the tweets.df data frame to the file airteltweets.csv. You can verify the data available in this file using Notepad++ or Excel In the next post we will look at how to do sentiment analysis with this file data.

Sentiment Analysis on Twitter data using R: Part – I

Now, the past posts we have understood the importance of using Twitter API, Basics of Twitter API and how we can access the Twitter API using R. Now we will get into analytics of how to do sentiment analysis with R with the library TwitteR. Before we do that we will try to do little understanding of Sentiment Analysis(some times also called as opinion mining) in a Q & A Format.

What is Sentiment Analysis ?

In simple words, Sentiment analysis is the task of identifying whether the opinion expressed in a text is positive or negative in general about a particular topic or context.

Can we have some examples?

a. I’m in a happy mood today, I go to beach. – Positive

b. I very much like R and its capabilities – Positive

c. I don’t like SPSS, its very complex to use – Negative

d. I feel Rapid Miner is easy to use and has good interface – Positive

Where it is being used or what are its applications?

With the lot of micro-blogging platforms available and business are well placed there, it’s important to understand that sentiment analysis on those platforms help understand the problems and feel of the customers.

  • Understanding customer feedback received
  • To arrive at happiness index of the customers
  • Determining product recommendations
  • Predicting Stock market moods
  • As a competitive marketing tool

Steps for Sentiment Analysis?

How do we do the Practical implementation?

http://www.slideshare.net/jeffreybreen/r-by-example-mining-twitter-for (The one I Like the most).

http://stackoverflow.com/questions/10233087/sentiment-analysis-using-r

https://sites.google.com/site/miningtwitter/questions/sentiment

http://trestletechnology.net/2011/11/sermon-sentiment-analysis/

You can look at the above given references for the practical implementation of Sentiment Analysis. Some of them may be outdated, in the next post we will do a practical step by step implementation of Sentiment Analysis with Twitter data using R.