I have data that is of the following form: https://imgur.com/a/Uq2sh B, C, and D are columns that contain repeated measures over N years in M cities. I am trying to use weka to classify/predict with SMO, the values of column B as a function of columns C and D (and maybe A and Year. not sure yet). I loaded the data in weka and I transformed column B into a 'nominal' column. Then I went to the classification tab, selected SMO, chose (Nom) B, and ran the code. I got the results posted below. I don't quite understand why the error measures are so high. And am I capturing correctly the fact that I am doing repeated measures in the same areas or do I need to specify something else to weka?
Correctly Classified Instances 80 %
Incorrectly Classified Instances 20 %
Kappa statistic 0.6374
Mean absolute error 0.2576
Root mean squared error 0.3308
Relative absolute error 85.2601 %
Root relative squared error 85.2565 %
Total Number of Instances 465
=== Detailed Accuracy By Class ===


TP Rate FP Rate Precision Recall F-Measure MCC ROC Area PRC Area Class
0.732 0.020 0.867 0.732 0.794 0.764 0.918 0.708 1
0.931 0.358 0.865 0.931 0.897 0.612 0.786 0.855 2
0.540 0.037 0.694 0.540 0.607 0.560 0.881 0.500 3
Weighted Avg. 0.847 0.263 0.842 0.847 0.842 0.628 0.819 0.784

=== Confusion Matrix ===
a b c <-- classified as
52 19 0 | a = 1
8 308 15 | b = 2
0 29 34 | c = 3