Benchmark 2e92831cb502

Parameter	Value
Benchmark hash	2e92831cb502
Original paper	https://doi.org/10.48550/arXiv.2503.17410
Dataset	CESNET-TimeSeries24
Aggregation	AGG_1_HOUR
Source	IP_ADDRESSES_SAMPLE
Type	Time-Based
Train size	0.35
Val size	0.05
Test size	0.6
Uni/Multi variate	Univariate
Metrics	n_bytes
Default value	0
Filler	None
Transformer	MinMaxScaler
Anomaly handler	None
Sliding window train	24
Sliding window prediction	1
Sliding window step	1
Set shared size	24
TS IDs	1

Related work	Model	Average RMSE	Std RMSE	Average R2-score	Std R2-score
https://arxiv.org/abs/2503.17410	GRU	0.149	0.82	-0.46	1.9
https://arxiv.org/abs/2503.17410	GRU_FCN	0.15	0.82	-0.12	1.1
https://arxiv.org/abs/2503.17410	INCEPTIONTIME	0.165	0.82	-2.7	3.9
https://arxiv.org/abs/2503.17410	LSTM	0.15	0.82	-0.41	1.8
https://arxiv.org/abs/2503.17410	LSTM_FCN	0.151	0.82	-0.44	1.9
https://arxiv.org/abs/2503.17410	MEAN	1.01	2.86	0.0	0.1
https://arxiv.org/abs/2503.17410	RCLSTM	0.221	1.08	-0.09	1.0
https://arxiv.org/abs/2503.17410	RESNET	0.152	0.82	-0.81	2.4