The splice site scores are computed with MaxEntScan (http://genes.mit.edu/burgelab/maxent/Xmaxentscan_scoreseq.html ).
A comparison of the scores can give a hint which splice site is preferentially used. The splice site with the higher score is often covered by more ESTs.
The score 'NaN' (not-a-number) means that the splice site context contains characters other than A, C, G, or T.
Nomenclature for the splice sites and transcripts
Tandem alternative splicing using the intron-proximal site (E acceptor / e donor - the distal part of the tandem becomes EXONIC) results in the E/e transcript, whereas splicing using the intron-distal site (I acceptor / i donor - as the entire tandem becomes INTRONIC) in the I/i transcript.
NOTE: MaxEntScan was trained on human data. Thus, the splice site scores for species other than human might have less meaning. However, since the splice site nucleotide preferences are quite similar between species (at least in vertebrates), we believe that scores for other species are useful as well.
 Yeo G & C Burge (2004) Maximum entropy modeling of short sequence motifs with applications to RNA splicing signals. J Comput Biol 11 (2-3) 377-394. PubMed