How to solve insufficient capabilities where it does not support polynominal attributes
hello, i am new to rapid miner, I'm trying to evaluate sentiment analysis (Neutral, Positive,Negative) but out all model i try only naive bayes worked, others show this error message "insufficient capabilities where it does not support polynomial attributes". why is this happening and what other models i can try?
Find more posts tagged with
Sort by:
1 - 2 of
21
Hi!
Your problem is not specific to text mining. It is a basic distinction between two-class and multi-class classification.
Many learning algorithms like logistic regression or support vector machines only support two-class (e. g. yes-no) problems. Here you're trying to predict three classes.
Naive Bayes, Decision Trees and other tree based methods, Neural Networks support multiple classes, but not all of them are good in text classification.
Another approach could be changing the label (target attribute) to a numeric score (e. g. -1 for negative, 0 for neutral, 1 for positive), building models that predict this score, and then finding good thresholds for deriving the classes from the numeric prediction.
Regards,
Balázs
Your problem is not specific to text mining. It is a basic distinction between two-class and multi-class classification.
Many learning algorithms like logistic regression or support vector machines only support two-class (e. g. yes-no) problems. Here you're trying to predict three classes.
Naive Bayes, Decision Trees and other tree based methods, Neural Networks support multiple classes, but not all of them are good in text classification.
Another approach could be changing the label (target attribute) to a numeric score (e. g. -1 for negative, 0 for neutral, 1 for positive), building models that predict this score, and then finding good thresholds for deriving the classes from the numeric prediction.
Regards,
Balázs
Sort by:
1 - 1 of
11
Hi!
Your problem is not specific to text mining. It is a basic distinction between two-class and multi-class classification.
Many learning algorithms like logistic regression or support vector machines only support two-class (e. g. yes-no) problems. Here you're trying to predict three classes.
Naive Bayes, Decision Trees and other tree based methods, Neural Networks support multiple classes, but not all of them are good in text classification.
Another approach could be changing the label (target attribute) to a numeric score (e. g. -1 for negative, 0 for neutral, 1 for positive), building models that predict this score, and then finding good thresholds for deriving the classes from the numeric prediction.
Regards,
Balázs
Your problem is not specific to text mining. It is a basic distinction between two-class and multi-class classification.
Many learning algorithms like logistic regression or support vector machines only support two-class (e. g. yes-no) problems. Here you're trying to predict three classes.
Naive Bayes, Decision Trees and other tree based methods, Neural Networks support multiple classes, but not all of them are good in text classification.
Another approach could be changing the label (target attribute) to a numeric score (e. g. -1 for negative, 0 for neutral, 1 for positive), building models that predict this score, and then finding good thresholds for deriving the classes from the numeric prediction.
Regards,
Balázs
please check the Rapidminer academy videos for Text processing and sentiment analysis.
https://academy.rapidminer.com/learn/course/text-and-web-mining-with-rapidminer
For sentiment analysis you need to convert your polynominal atributes to text and then process your data.
Best
Cesar