## Visitor's Comments

## Visitors' Comments on Tutorial: Similarity

We have 17 comments on this tutorial

Answer > How to calculate covariance and inverse pooled covariance and means difference from mahalanobis distance tutorial (date: 2010-10-23)

By Kardi

Hi Noor, You may want to download the excel file companion of Mahalanobis tutorial. Its located at the bottom page of http://people.revoledu.com/kardi/tutorial/Similarity/MahalanobisDistance.html

Thank you > Nice Post (date: 2010-12-10)

By Richard

Just exactly what I was looking for! Concise, directional and quite helpful.

Question > Statistical Distribution (date: 2010-12-14)

By Md. Ali Hossain

Hi, This is Md. Ali Hossain. I am a PhD Student. I am working with a large array of images (like 220 band images). After applying PCA I have got some real and negative value for intensity levels. Now I want to normalize intensity levels for Principal components1 to Principal components220 and wants to take their variance in order from higher to lower for all the 220 components. In that case which normalization process will be helpful. I am using normalization with maximum values and causing later components showing higher variance that is a problem. I want to get a varaiance from higher to lower for pc1 to pc22(PC: principal components).

Question > How to convert the mahalanobis distance into percentage (date: 2011-01-05)

By noor aznimah

hi, based on your tutorial, how can we convert the mahalanobis distance results to percentage. i already check the z-score but need more clarification on this. my input data is exactly as your tutorial, really appreciate if you could elaborate more. thank you so much

Question > Canberra Distance (date: 2011-01-05)

By Chuan

Can you check for your Canberra distance formula?If no wrong, the range of Canberra always fall in range 0 to 1. Here I get a question:How I interpretation for the value for Canberra Distance, is it more close to 1, it means the distance between to coordinates are similar and vice versa?

Answer > Canberra Distance (date: 2011-01-06)

By Kardi

The value of Canberra is distance is not 0 to 1. The formula is correct. Each term of the fraction difference has value between 0 and 1. You may consider Bray Curtis (Sorensen) distance if you need a normalized distance.

Answer > Percentage (date: 2011-01-06)

By Kardi

Percentage or ratio should be measured based on the maximum or the range. Please check my Normalization tutorial (http://people.revoledu.com/kardi/tutorial/Similarity/Normalization.html)

Question > help on statistical normalization tutorial for mahalanobis distannce (date: 2011-01-07)

By noor aznimah

Hi, can you show the tutorial on statistical normalization for mahalanobis distance to get the percentage. i have data that would return vertex points of shape (coordinate x and y) example, group A :(x1,y1), (x2,y2), (x3,y3), (x4,y4),(x1,y1), (x2,y2) and group B: (x5,y5),(x6,y6). my data group would have two group, group A (39 coordinates) and group B (4 coordinates). i applied mahalanobis distance to calculate the similarity computation and mahalanobis distance return standard deviation results from expected data rights. please correct if im wrong. my questions: 1) how im going to get the accuray (percentage) of the mahalanobis distance and what method (conversion the mahalanobis (standard deviation to percentage)) is better use for my data? for now, as my review, i need to look up at z-score table, which means to calculate by z=(X-mu)/standard deviation. i already trying statistical normalization but getting confuse to place the input data for X, u and standard deviation based on my data really appreciate if you could share knowledge and ideas or computation on this thank you so much

Answer > help on statistical normalization tutorial for mahalanobis distannce (date: 2011-01-07)

By Kardi

From a single data point you cannot do normalization, unless you know the maximum value or the range of min to max. I added an example for statistical normalization in http://people.revoledu.com/kardi/tutorial/Similarity/Normalization.html

Thank you > K mean clustering (date: 2011-11-29)

By Mohsin Khan

Than q, i got a good idea about K means clustering

Thank you > I frequently refer your tutorials to my students (date: 2013-11-01)

By Badar Sami

It is one of the most comprehensive and easy to understand tutorial that I ever studied on this topic. I frequently refer this and other tutorials on your site to students. I referred this to students while teaching Text mining and Natural Language Processing to CS major undergrad students in their final semester.

Question > it sucks dude nothing is understandable (date: 2015-11-18)

By Furkan Gozukara

it sucks dude nothing is understandable here a better question if you answer i appreciate https://www.quora.com/How-Do-I-calculate-Mahalanobis-distance-at-c-working-example-tutorial http://stackoverflow.com/questions/33778968/how-to-calculate-mahalanobis-distance-at-c-sharp-working-example-tutorial

Thank you > machine learning for kmeans and distance calculation (date: 2015-12-01)

By kazuki

good!

Thank you > datamining (date: 2015-12-29)

By gangappa

good

Thank you > Great tutorial for learners and seekers (date: 2016-01-17)

By Rupak Aryal

I am looking your tutorial and there are lots to learn for me. Thank you for your great effort to make math/statistics so simple

Thank you > Canberra distance (date: 2016-02-28)

By Eduardo m a peixoto

Simple and clear.

> (date: )

By