Im planning to make a 2 courses data mining for bachelor students.. the couses are title 1- clustering algorithms in data mining 2- classification data mining.
Im listing here the algorithms ill demonstrate..
The question is: which of these algoritms are considered to be clustering or classification?
Algorithms:
1- association rule
2- apriori
3- naive baysian
4- regression
5- k-means
6- tree id3
Thanks
Put all the above algorithms in both scenarios except id3 which falls in classification only.
The basic idea is that clustering and classification differ in many concepts such basic knowledge of the classes (in clustering this criteria is required while in clustering isn't), also classification classifies samples into a specific known classes wile clustering classifies them into suggested groups. However, K-means, Kprototypes, Fuzzy C Means, Hierarchical Clustering are the most common clustering algorithms in Data Mining, While naive Bayesian, Logistic Regression, Linear Regression, Trees, association rules are commonly used as a classification methods. The last thing is that you can use the output of clustering algorithms (such as K-means) as inputs of the classification algorithms (here we called these inputs as features).
Tareef Kamil Mustafa,
Here are all the algorithms you listed and their categories:
Association rule mining:
- association rule (this is the category name--not the technique)
- apriori
Cluster:
- k-means
Classification:
- Naive Bayes
- regression (Multinomial logistic regression)
- tree id3
Regression
- regression (Linear Regression)
HTH.
Dr. Samer Sarsam
I wouldn't categorize the different learning methods that you've listed as either classification or clustering because some of them are neither. Classification refers to methods of supervised learning that deal with data that has a categorical target variable. Clustering refers to methods of unsupervised learning that discovers groups in data that does not have a target variable. So your question should be if these messages that you've listed are supervised or unsupervised learning methods.
Methods that deal with data with no dependant variable are unsupervised learning methods. This includes clustering and association rule mining. Methods that deal with data that has dependant variable are supervised learning methods. This includes classification and regression.
1- association rule: unsupervised
2- apriori: unsupervised
3- naive baysian: supervised
4- regression: supervised
5- k-means: unsupervised
6- tree id3: supervised
Hi again,
Just wanted to highlight one issue that seems to be necessary for this discussion.
Khobaib Zaamout, many books contain independent sections that are named as "Classification", Cluster" or "Cluster analysis", and "Association rules". Therefore, I don't see any issue with the terms "Classification", "Cluster", and "Association rules" to be used in the course of Tareef Kamil Mustafa. Hence, the terms "Supervised learning" and "Unsupervised learning" would not really change the core of the course, but rather, such terms could increase the confusion of the learners. For example, when saying unsupervised learning, the student will not easily recognize whether this technique is a cluster or an association rule mining which are two different techniques.
Tareef Kamil Mustafa, if the course is mainly about technical aspects in classification or cluster or association rules, then after you know the type of each of the listed algorithms from my previous post or colleagues' posts, you can explain the algorithm/its type (supervised/unsupervised), and then its technical features.
Dr. Samer Sarsam
Dr sameer
I Apreseate your Positive reply.
We are limited in academic terms, i can use a course titled 'data mining' and there general diversions such as cluster, classification.. rather then a cose name 'unsupervised'.. its bachelor course and we must be on basics..
Dear Tareef Kamil Mustafa ,
You are most welcome.
If the current issue is mainly the title of the provided course, then you can call it "Introduction to data mining" or "Data mining course for beginners" or "Data mining: Basic concepts". Then, if necessary, you may also highlight the content of the course, e.g., classification, cluster, etc.
Dr. Samer Sarsam
- k-means clustering
- Naive classification,
- Tree id3 classification
Association rule is a simple clustering algorithm because it generate rules rather than predict them.
K-means is one of a famous clustering algorithms.
Naive Bayesian, id3 and regression are classification algorithms.
Mustafa S. Abd, appreciate your contribution, but I would like to correct your answer.
Association rules are not quite different from classification rules except that they can predict any attribute, not just the class, and this gives them the freedom to predict combinations of attributes too. Thus, association rule algorithm is NOT a clustering algorithm as you claimed.
Additional issue needs to be clarified; if the regression method is "Multinomial logistic regression", then it is considered as a "Classification" technique. In contrast, if the algorithm is "Linear Regression", then it is a "Regression" technique. This means that classification and regression are not same.
Dr. Samer Sarsam
Association rules method is using usually for finding the frequent items within a dataset, the common application of this kind of analysis is the basket market analysis, to find items or set of products bought together by a specific customer for many purchase trips.
Also, to find out the frequent items, which usually purchased together. there are many algorithms for that like the Apriori Algorithm.
Hi,
How you intended to hold this course, online ???
Why don't you star towards Deep learning?
Thanks
@Hiba J. Aleqabie
The courses will be designed as slides given in college
@ Tareef Kamil Mustafa
can you, gratefully, Send me an invitation to join the course.
- Badges
- Science topic
- Similar topics
- Computer Science
- Data Mining