High dimensional clustering input importance
1 vue (au cours des 30 derniers jours)
Afficher commentaires plus anciens
Hello,
I am venturing in to new territory and thought I would seek a little guidance.
I am looking at data retroactively to try to determine input importance relative to a known output. Lets say I have X input parameters, I am trying to determine a range to individually filter input parameters such that
for i=1:n Y(i) = find (x(i) > X(i)min && x(i) < X(i)max ) end
whereby Yi:Yn maximizes the number of X input parameters relative to a classifcation (true or false).
In perhaps more simple to communicate terms. I have marketing survey data for 1000 individuals that involves 10 questions that are bound to a range -100 to 100. Assume that 100 individuals answered 'Yes', and another 100 individuals answered 'No',* I am trying to find a range for answers to the 10 questions that is most likely to produce a yes or a no.* I then want to use this range to filter out current data to target a search.
I am considering kmeans clustering to find the largest cluster groups and looking at the distribution of inputs to determine a range. Another thought was SOFM to get a map and then look at the neurons with the most hits and then also implement a distribution of inputs to determine a range.
Thanks Very much for any feedback.
Beav
0 commentaires
Réponses (0)
Voir également
Catégories
En savoir plus sur Cluster Analysis and Anomaly Detection dans Help Center et File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!