Using Maps with rapid miner

supadhyaya

HI ,

I have a table which contains userIDs and their IP address. I want to display the density of user according to the region where they are loggin in from using a MAP.

can anybody tell me if rapid miner has digrams / plots which will let me visualize the density using the MAP (something like map of the world)

Regards,
supadhyaya

BalazsBarany

Hi!

RapidMiner has no ready-to-use functions for this.

You could geocode the IP addresses using the Maxmind GeoIP database: http://dev.maxmind.com/geoip/geoip2/geolite2/

For map display, you always need to select a "map projection" and transform the coordinates accordingly.

Then, using the reporting operator, you export a scatterplot with the transformed coordinates. This plot could be laid over a base map.

It's quite an involved process, not really easy.

I'd do the display part using an online map server software like OpenLayers http://www.openlayers.org, Leaflet http://leafletjs.com/ or an offline map generator like MapServer's shp2img tool: http://mapserver.org/en/utilities/shp2img.html.

Rene

Yes, one could use the GeoIP database to recode IPs into longitude & latitude or country data. Then I'd suggest you to use RapidMiner's R extension. R has some nifty functions to plot geo data with no effort.

supadhyaya

Thank you guys,
Yes I also think that the process you mentioned is really involving but surely is worth trying... Once again I thank for your advice. I am taking it and working on it currently..

Cheers

wilsonchua

You could try the community edition of SPLUNK with the google plugin to do what you want to achieve.

MBM

So for example when I have lat and long as points on a Map in RM usinge R extension can I use the points that are displayed for clustering?

Thomas_Ott

Yes, you can use the points that are displayed for clustering provided you write your R code to pass that data back to a RapidMiner process. 

 

Something like this:

rm_main=function(data)

{

library maplibrary

 

some R code to generate your stuff

 

data <- mylatlongforclustering

 

return (data)

}