Machine learning on wireless spectrum data

Hi,

I am having wireless data from spectrum analyzer.

I am having file in h5 format consists of values in dBm.

I would like to use unsupervised learning technique on this data.

My requirement is to cluster points having certain value for some period of time like if signal level is let say -70dBm for more than an hour then i will cluster those points as high priority. Another classification can be based on occupancy like if occupancy is more than 80% for longer duration then priority is high.

Please guide me on how should i use machine learning tools to cluster points and number of classes i could find out based on these values.

Regards Vaishali