In the IRIS dataset (attached), I test with every method like LSVM, QSVM,NARROW NEURAL NETWORK, and WIDE NEURAL NETWORK. For data numbers 71 and 84, the answer is wrong. Could this data be wrong?
not all the iris database get accurate results, change the type of database
Seyed Hossein Ahmadi ,
You should download the dataset from UCI
https://archive.ics.uci.edu/dataset/53/iris
There it states there that:
"One class is linearly separable from the other 2; the latter are NOT linearly separable from each other"
It is also worth reading where the data came from and how it was collected. For more information read the introductory paper there which is:
The Iris data set: In search of the source of virginica
By A. Unwin, K. Kleinman. 2021
Also the original paper by Fisher can be foundin [1]. I would also get the Anderson papers given in Kleinman before making a determination if it is really an error or not.
References
[1] THE USE OF MULTIPLE MEASUREMENTS IN TAXONOMIC PROBLEMS BY R. A. FISHER
Available at(https://onlinelibrary.wiley.com/doi/pdf/10.1111/j.1469-1809.1936.tb02137.x)
Hi, I saw both the UCI website and Fisher's original article, which includes both mode numbers 71 and 84 (5.9, 3.2, 4.8, 1.8) , (6, 2.7, 5.1, 1.6) as part of the versicolor type while with every artificial intelligence method that I check, these values are part of the virginica Does anyone have a justification for this?
I read Fisher's original article, which claimed that this method has an accuracy of 3 errors per million, but at the end of the article, he mentioned that the species versicolor is genetically a hybrid species of virginica so we can guess why I have assumed that there was a mistake because it is possible that the two species are genetically very similar.
Anyway, thank you for your cooperation.
- Badges
- Science topic
- Similar topics
- Condensed Matter Physics
- Zero