Experiment with incremental amount of blocks and compare with the baseline

[sukruc]

Results for different numbers of blocks:

• 1 block
  • Best RMSE is [4.550956, 5.4067826, 5.884017]
  • Best MAE is [2.2905083, 2.5936992, 2.7025142]
  • MAPE: [0.21264371, 0.23939687, 0.25425482]
• 2 blocks
  • Best RMSE is [4.5192447, 5.3059397, 6.055376]
  • Best MAE is [2.238572, 2.5285902, 2.7109706]
  • MAPE: [0.20852102, 0.22923572, 0.2518258]
• 3 blocks
  • Best RMSE is [4.490098, 5.254158, 6.227891]
  • Best MAE is [2.2533786, 2.5316834, 2.7846472]
  • MAPE: [0.2091166, 0.23115602, 0.25449482]
• 4 blocks
  • Best RMSE is [4.4959836, 5.251608, 5.982039]
  • Best MAE is [2.2391803, 2.4735162, 2.7636094]
  • MAPE: [0.2098142, 0.23014443, 0.25120753]
• 5 blocks
  • Best RMSE is [4.50711, 5.3540206, 5.9957166]
  • Best MAE is [2.2466822, 2.539691, 2.7361522]
  • MAPE: [0.20992851, 0.23677891, 0.25073275]
• 6 blocks
  • Best RMSE is [4.5349536, 5.5062766, 5.900797]
  • Best MAE is [2.2975507, 2.5579286, 2.6789894]
  • MAPE: [0.20868744, 0.232984, 0.24766727]

[sukruc]

Full training output

nik-results.txt

TF Summary: onedrive link