Accuracy stops increasing at around 82%

I was wondering how I should weight the data as my accuracy stops at around 82%.

Test Classes Actual: [138, 156, 407, 450, 9334, 2192, 12087, 987, 828]
Test Classes Classified by Model: [81, 16, 252, 12, 9069, 0, 15339, 982, 0]

How should I weight the values, because when I make the weights all equal, I get this?

Test Classes Classified by Model: [430, 141, 934, 912, 6619, 7395, 8332, 1050, 0]

Any help would be appreciated.

Thanks,
Rishav