0
$\begingroup$

After performing some sentiment analysis, I have a dataset that looks like this: For different products, using online reviews, I have obtained some values for positive/negative sentiments. However, now I am unable to figure out how to draw conclusions for this.

I had the idea of using correlation but need ideas on what features could be created & what comparisons could be made?

The dataset includes different "Features" like webcam, screen, mousepad for different products (product name).

id     Date       Website       Product Name    Brand   Stars   Feature   Sentiment  Positive     Negative       Anger      Happiness       Annoyance
0     2020.8.03   eBay    Lenovo Hi-Fi 320    Lenovo     4      Screen      NEGATIVE    0.000047    0.999851    0.000101    0.108132    0.248220    
$\endgroup$
0
$\begingroup$

Depends what your Goal is.

But generally you see that positive and negative Sentiment probabilities are disjunct. Meaning your model focuses for one class only, negative one. ANd thats it, you conclude your data sample belongs to class "Negative". What that means, depends on the Definition of this class.

$\endgroup$
1
  • $\begingroup$ I have only copied one row from my data. There are different websites, different products, different sentiments (positive/negative) both. The goal is to conclude any useful information that I can but for now, I am struggling with how I can relate the variables. More like which variables should I compare with what variables and what info I can extract using this limited data. Vague question, I know. sorry $\endgroup$ – x89 Nov 8 '20 at 12:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.