If we make clusters for patterns but each pattern was not equal in numbers of position in playground because the time for each one is different so sometimes. we have 200 position in coordinates x and y and some others we have only 150
Sport neural network
Dear Amr Abdulfattah Hassan
in case your data is not uniform, you can try to "fill in " the missing data using assumptions acquired from experience, theory or by statistical methods (see answer from Ashish Dutt)
Another method is to extract features from the original data sets and perform the clustering using these features. Typically these features are chosen, or constructed in such a way, that they are invariant with respect to some aspects. e.g. rotational invariance, or scaling invariance or ...
For your specific case it would be necessary to choose features that are invariant with respect to the number of values in each pattern.
Sínce i do not know the application you have in mind, and thus i do not know what these patterns are representing, it is a bit difficult to propose features.
See the method of Bag Of Words , or Bag of Visual Words.
A nice introduction can be found:
http://people.csail.mit.edu/torralba/shortCourseRLOC/
There are two ways:
1) Fill the missing values. There are many approaches such as using global constant, use mean value, or use more probable value[Data mining : Concept and Techniques, Jiawei Han &Micheline Kamber . page 109] or [Handling Missing Values in Data Mining, Bhavik Doshi].
2) Reduce the extra values from x by using mean or median for some values that you want to remove.
- Badges
- Science topic