XLSTAT: Sentiment Analysis

Dataset for Example

The attached dataset contain the lyrics of 94 American singer Taylor Swift songs.

The dataset is extracted from the data science platform, Kaggle and might be accessed at this address.

 

Setting Up

• Open XLSTAT.

•  Click on the Text Mining menu and select the Sentiment analysis feature.

• In the General tab of the dialog box:

  • Select the term frequencies from the document-term matrix.

  • Choose the Bing sentiment dictionary for binary sentiment classification (positive/negative).

  • Select song titles as document labels.

• Click on the OK button to launch the computations.

• The results are displayed on a new Excel sheet.

 

Interpreting the Results

The document score is the sum of the products of the term score multiplied by the term frequency. The more positive a document is, the higher its sentiment score. In our case, the song with the most positive intent is "Out of the woods" while the most negative is "Shake it off".

The distribution of the scores of the documents allows us to know here that the majority of the songs have a score close to zero, thus neutral. 

The sentiment-based word cloud is interpreted as a classic word cloud but here the words are colored according to their sentiment. It shows that "love" is the most present positive word in the 94 songs of Taylor Swift, while "shake" and "bad" seem to be the most present negative words. This explains why the song "Shake it off" is categorized as negative. 

Tips: The score of the term "shake" can be decreased or neutralized so that it does not influence the score of the song "Shake it off". To do so, you must activate the custom scores option.
The last chart gives a view of the scores of each term, it allows seeing the sentimental category of each term.