Data Mining and Knowledge Discovery

Follow this topic

Answer added to:
2 I'm looking for papers on computational geometry-based approaches to data clustering
By Colin Groves · University of Idaho

Nam Nguyen V.

Simple Spatial Clustering Algorithm Based on R-tree http://link.springer.com/chapter/10.1007%2F978-3-642-35455-7_22

Simple Spatial Clustering Algorithm Based on R-tree http://link.springer.com/chapter/10.1007%2F978-3-642-35455-7_22

Following
Answer added to:
1 Can anybody suggest which algorithm is good for outlier analysis in Datamining?
By Savita Shetty · M.S. Ramaiah Institute of Technology

Sergey Antipov · Moscow Power Engineering Institute (Technical University)

It's hard to answer your question without knowing at least the domain you're working at, so you may have a look at the following article, which provi... [more]

It's hard to answer your question without knowing at least the domain you're working at, so you may have a look at the following article, which provides overview of existing outlier detection techniques by classifying them along different dimensions: VARUN CHANDOLA, ARINDAM BANERJEE, VIPIN KUMAR "Outlier Detection : A Survey" http://www.bradblock.com.s3-website-us-west-1.amazonaws.com/Outlier_Detection_A_Survey.pdf (or just google this article) You'll probably find what you need following the references.

Following
Answer added to:
2 What are the relations among the three terms uncertainty, impreciseness and vagueness?
By B.K. Tripathy · Vellore Institute of Technology

Russell Thomas · George Mason University

"uncertainty" has plural meanings, depending on context and discipline. It is really an umbrella term that covers several semi-related concepts assoc... [more]

"uncertainty" has plural meanings, depending on context and discipline. It is really an umbrella term that covers several semi-related concepts associated with "lack of certainty" in reasoning. Varieties of uncertainty include vagueness, variability, imprecision, indeterminacy, undefined, unknowable, absence (i.e. missing information), error, and so on. "Imprecise" has a more specific meaning than "uncertain". In numerical contexts, it basically means fewer significant digits. If the true value is "3.14159...." and the estimates are "3" vs. "3.14", then the first is imprecise compared to the second. Similar distinctions can be made in qualitative/structural/linguistic domains. "Vague" combines imprecision with generality and corresponds to a range or region of values, and roughly corresponds to "fuzzy specification" or "fuzzy estimation". The adverb "about" is often used in vague expressions, i.e. "He is about 6 feet tall". (Notice we never hear "He is about 6.231193 feet tall".) For more, see these books by Michael Smithson: Uncertainty and Risk - http://www.amazon.com/dp/1844078515/ Ignorance and Uncertainty - http://www.amazon.com/dp/0387969454/

Following
Answer added to:
2 I am looking for a recent state of the art in using NLP to identify linguistic markers of argumentation in dialogs (written text as input).
By Frédérique Segond · Viseo Research Centre

Chandra Amaravadi · Western Illinois University

Some related work has been done at Simon Fraser U (I don't work there). http://www.sfu.ca/~mtaboada/research/pubs.html

Some related work has been done at Simon Fraser U (I don't work there). http://www.sfu.ca/~mtaboada/research/pubs.html

Following
Answer added to:
2 Is there any research done on the intention of mining in text analytics?
By Shailja gulati · Tata Consultancy Services Limited

Following
Answer added to:
1 Are there any sites with a suitable page that has an abundant amount of links for different works on genetic data and literature?
By Aleksei Sleptcov · Research Institute of Medical Genetics

Deborah Hilton · Deborah Hilton Statistics Online

Helo Dr Sleptcov, I am in Australia and this is not my area of expertise but I did hear this website mentioned so you could email someone from this gr... [more]

Helo Dr Sleptcov, I am in Australia and this is not my area of expertise but I did hear this website mentioned so you could email someone from this group. http://agrf.org.au/ Thanks Deborah

Following
Answer added to:
2 What is the effect of eliminating some fixed attributes from all datasets on classification?
By Pouya Sinaian · Blekinge Institute of Technology

Pouya Sinaian · Blekinge Institute of Technology

@Marco Thank you. It was helpful. So, do you think elimination of fixed attributes from a dataset has no negative effect on classification?

@Marco Thank you. It was helpful. So, do you think elimination of fixed attributes from a dataset has no negative effect on classification?

Following
Answer added to:
6 A data set in SNA with structural (link properties) and also demographic, geographic or some more features of users?
By Afarin Adami · Islamic Azad University of Najafabad

Afarin Adami · Islamic Azad University of Najafabad

@VIad thanks alot it was exactly the thing i was looking for.

@VIad thanks alot it was exactly the thing i was looking for.

Following
Answer added to:
44 Statistical method for dimension reduction
By Maen Alaraj · Yamagata University

Bruce Carnes · University of Oklahoma Health Sciences Center

Oh, forgot. It is easy to determine the dimensionality of your data. However, one question that has to be addressed is whether your new composite vari... [more]

Oh, forgot. It is easy to determine the dimensionality of your data. However, one question that has to be addressed is whether your new composite variables remain correlated (Factor Analysis) or are independent (PCA). Also, the biggest issue is what do these new dimensions mean. You can create them, but what are they? The same problem exists with recursive partitioning. As far as I know, there is no quantitative solution to this "interpretation" problem.

Following
Question:
Open Which is the best implementation of SVM on a GPU available on the web?
Have you tried any of the GPU based SVM implentations available online? Were you satisfied with the speedup you obtained w.r.t. a traditional CPU base... [more]

Have you tried any of the GPU based SVM implentations available online? Were you satisfied with the speedup you obtained w.r.t. a traditional CPU based implementation? Please state the characteristic of the dataset you used in your trials (number of features, number of samples, etc...)

By Marco Trincavelli · Örebro universitet

Following
Answer added to:
3 When sub graph patterns are mined, how are graphs efficiently kept in memory?
By Maryam Taki · Islamic Azad University of Najafabad

Following
Answer added to:
4 Who is doing research on computational predictive models of health behavior change?
By Peter Pirolli · Palo Alto Research Center

Stuart Card · Stanford University

These are weak candidates for models that are computational, but they do have grist for such models. Consolvo, Sunny; McDonald, David W.; CHen, Mike... [more]

These are weak candidates for models that are computational, but they do have grist for such models. Consolvo, Sunny; McDonald, David W.; CHen, Mike Y; Froehlich, Jon; Harrison, Bevery; Klasnja, Predrag; LaMarca, Antony; LeGrand, Louis; Smith, Ian, & Landay, James A. (2008). Activity sensing in the wild: A field trial of UbiFit Garden. CHI 2008. Marlow, Peter (2005). Literature review of the value of target setting. Technical Report HSL/2005/40, Health and Safety Laboratory, Broad Lane, Sheffield. Duhigg, Chalres (2012). The Power of Habit. New York: Random House. Heath, Chip & Heath, Dan (2010). Switch: How to change things when change is hard. Crown Business. --Stu

Following
Answer added to:
7 What are pros and cons of decision tree versus other classifier as KNN,SVM,NN?
By Mahdieh Najafy · Zanjan University

Ali Fakhari · Qazvin Islamic Azad University

for the type of input data, i can say that CART algorithm can handle continuous data and of course can be used in regression problems. These are pros ... [more]

for the type of input data, i can say that CART algorithm can handle continuous data and of course can be used in regression problems. These are pros of DT: + Ability of selecting the most discriminatory features. + Comprehensibility so that can be used in Rule Generation problem + Data classification without much calculations + Dealing with noisy or incomplete data + Handling both continuous and discrete data (you have to choose proper algorithm) Cons: - The high classification error rate while training set is small in comparison with the number of classes - Exponential calculation growth while problem is getting bigger. - Need to discrete data for some particular construction algorithm.

Following
Question:
Open Availabe techniques for user profiling, and for recommender systems from similar profiles.
Scenario: We have an application with several thousand of user options. Each user selects a set of options from time to time (low frequency). So each ... [more]

Scenario: We have an application with several thousand of user options. Each user selects a set of options from time to time (low frequency). So each user might have an history of previous sets of selected options. Also each time a user is going to make a new selection of options, we might pre-select some options in a deterministic fashion (due to external information sources). Some options are related to others. There exist a manual classification of options, with relationships between them, which also can be used to tag user characteristics. A user Profile would consist in a group of characteristics derived from external information, and additional options selected which belong to a class. Ranges: Options: Thousands. Options in User Sets: Max tens. Users: 1Million. Previous Sets: [0,30] Problem to solve: We want to help the user to build the right set of options each time it uses the application. So when a user starts a new selection process, we: 1. Might have deterministic selected options. 2. Might have previous selected sets, which have a certain probability of being re-selected (heuristics can be retrieved for this). 3. We plan to make a short questionnaire to determine if the user has several characteristics (tags) with a high probability of being true for that user in that moment (from data analysis of all users). 4. We should be able to recommend options to the user according to the new tags, and other options that might define the user (from previous data analysis). We do have big amounts of data of previous set selections, so that can be used to test the accuracy of the methods used. Question What techniques would you recommend for this scenario? Approaches, Languages, Algorithms, Databases, Frameworks, Papers that talk about similar situations, etc. .

By Javier Ruiz Aranguren · Independent Researcher

Following
Answer added to:
4 Data Mining to Reduce Poverty

Deleted

Thank you for the answers,... is there any special characteristic in the data such that current data mining methods can't perform optimally? I am look... [more]

Thank you for the answers,... is there any special characteristic in the data such that current data mining methods can't perform optimally? I am looking forward for challenges in the data such that a modification of the previous method is needed.

Following
Answer added to:
3 Specificity and Sensitivity
By Sreeraj m.s · Cochin University of Science and Technology

Alex Mitchell · University of Leicester

We have just launched an interactive online website to help anyone wanting to work out & interpret diagnostic tests for accuracy and clinical utility.... [more]

We have just launched an interactive online website to help anyone wanting to work out & interpret diagnostic tests for accuracy and clinical utility. The website is www.clinicalutility.co.uk. On that page go to the excel xls doc for best value

Following
Answer added to:
5 Problem in classifying 20 concepts.
By Numan Rauf · Quaid-i-Azam University

Adis Medic · Infosys

I think datamodeling is very helpfull, but give more instruction.

I think datamodeling is very helpfull, but give more instruction.

Following
Question:
Open Pseudo R squared acceptable values for Predicted vs Observed plot
Are there any articles that list acceptable rates for PRvsOB plots for models? I managed to push my model up to 32% (numerical predicted variable). I ... [more]

Are there any articles that list acceptable rates for PRvsOB plots for models? I managed to push my model up to 32% (numerical predicted variable). I can set an arbitary judgement but is there an exact guideline for this?

By Michael Tsikerdekis · Masaryk University

Following
Answer added to:
6 Help on reporting data mining models (Boosting, Support Vector Machines etc.)
By Michael Tsikerdekis · Masaryk University

Ronaldo Prati · Universidade Federal do ABC (UFABC)

What about PMML: http://en.wikipedia.org/wiki/Predictive_Model_Markup_Language Several softwares (weka, R, etc) can save and load models in PMML

What about PMML: http://en.wikipedia.org/wiki/Predictive_Model_Markup_Language Several softwares (weka, R, etc) can save and load models in PMML

Following
Answer added to:
8 What is the *concept* of class in your favorite learning model, e.g. in SVM, ANN, etc.?
By Lev Goldfarb · University of New Brunswick

Lev Goldfarb · University of New Brunswick

Antonio, You can sleep well for now. ;--) ETS is not coming yet for the 'public' consumption; it is for the ambitious young turks in science and engi... [more]

Antonio, You can sleep well for now. ;--) ETS is not coming yet for the 'public' consumption; it is for the ambitious young turks in science and engineering. However, I'm trying to get some investments for developing ETS-based search engine. Have fun!

Following
Answer added to:
11 How can we find out mathematically how complicated is a concept to train a machine with?
By Pouya Sinaian · Blekinge Institute of Technology

Pouya Sinaian · Blekinge Institute of Technology

thanks Ramon and Xavier

thanks Ramon and Xavier

Following
Answer added to:
10 I am just started doing research in Datamining concepts, can any one suggest how can we get the problem etc., please guide me.
By Sathiya Moorthy Srinivasan · Oracle Corporation

Hervé Nguendon · Leuphana University Lüneburg

Please spend a time on my publications and you will have more idea about datamining

Please spend a time on my publications and you will have more idea about datamining

Following
Answer added to:
5 How RG evaluates the impact of the conference papers?
By Nazar Zaki · United Arab Emirates University

Nazar Zaki · United Arab Emirates University

The idea of adding score to quality questions and answers is excellent idea however, I agree that publications are more important as one paper may con... [more]

The idea of adding score to quality questions and answers is excellent idea however, I agree that publications are more important as one paper may contribute to answering several valuable questions.

Following
Question:
Open What are utility measurements available in utility based privacy preserving methods?
What are utility measures available in Utility based privacy preserving methods?

What are utility measures available in Utility based privacy preserving methods?

By Mitali Sonar · Sardar Vallabhbhai National Institute of Technology

Following
Question:
Open Is anybody running RDF triplestores?
If so, which implementation and related tools do you find most useful?

If so, which implementation and related tools do you find most useful?

By Matej Lexa · Masaryk University

Following
Answer added to:
1 How can we verify the shares of secret in verifiable secret sharing scheme?
By Mitali Sonar · Sardar Vallabhbhai National Institute of Technology

Amit K Awasthi · Gautam Buddha University

Please read lagrange's interpolation (from any book of numerical methods/numerical analysis) then post messgae I will help you to understand this pro... [more]

Please read lagrange's interpolation (from any book of numerical methods/numerical analysis) then post messgae I will help you to understand this problem. Following literature may be helpful. P. Feldman, A practical scheme for non-interactive verifiable secret sharing. IEEE Symposium on Foundations of Computer Science, pages 427--437. IEEE, 1987. doi:10.1109/SFCS.1987.4 Oded Goldreich, Secure multi-party computation

Following
Answer added to:
6 Hybrid classifiers in classification mining
By Janjira Jojan · Kasetsart University

Following
Answer added to:
4 Which ANN is recommended and how should I best construct time series variables to predict as aspect of personal risk? Would a decision tree be better?
By Allan Brimicombe · University of East London

Pierre-Francois Marteau · Université de Bretagne Sud

Allan, here is my understanding (sorry if I am wrongly interpreting your purpose) : Your data are multidimensional time series of various lengths, non... [more]

Allan, here is my understanding (sorry if I am wrongly interpreting your purpose) : Your data are multidimensional time series of various lengths, non-uniformly sampled : Basically : X=x(t1)x(t2)...x(tk), where x(ti) \in S (basically R^n union some semantic set), and ti is the increasing time-stamp. So two individuals are represented by time series having different lengths that are potentially characterized with disjoint time-stamps. Distances and even metrics exist that can cope with such time series data, non-uniformly sampled and having various lengths. I call them elastic distances, they are typically based on dynamic time warping (DTW). My point is you can try to use such distances as a similarity measure usable in some machine learning algorithm. In this way you keep track of the historical information and would be able to detect typical time patterns. My first choice for the machine learning algorithm would be SVM, since elastic kernel based on elastic distances can be proposed for such machine. I guess It would be less straightforward to integrate such measures into ANN or DBNN (as suggested by Omar).

Following
Answer added to:
2 Which protocols are applied for privacy preserving supporting malicious models?
By Mitali Sonar · Sardar Vallabhbhai National Institute of Technology

Mitali Sonar · Sardar Vallabhbhai National Institute of Technology

These techniques are applied to privacy preserving.Here the participant in protocol comes under 2 categories 1)semi honest 2)Malicious..Most of the so... [more]

These techniques are applied to privacy preserving.Here the participant in protocol comes under 2 categories 1)semi honest 2)Malicious..Most of the solution to privacy preserving data mining assumes semi honest participants.So i want to know about techniques that are extended to support malicious models specifically that includes secret sharing key

Following

About Data Mining and Knowledge Discovery

It is the research project which is ongoing.

Topic Followers (6400) See all

Chih-Sheng Chen

Tzu Chi College of Technology
Mahmoud Abu Moustafa

Central University of Finance and Economics
Gregory Funaro

NASA
Sabir Ismail

Shahjalal University of Science and Technology
Kamlesh Kumar

Uttarakhand Technical University
David Markus

Saddleback Geosolutions
Chen Xiangping

Jiujiang University
Juan Ignacio Guerrero

Universidad de Sevilla
Sandra Servia-Rodríguez

Universidade de Vigo
Mike Denis

National Health Service

Data Mining and Knowledge Discovery

About Data Mining and Knowledge Discovery

Topic Followers (6400) See all

Log in