Examining parameter importance in modelling with mobility and weather data

Created on May 18|Last edited on June 13

Comment

﻿
 Italy region -- Lombardy (with only new cases)
﻿
﻿
﻿
Parameter importance with respect totest_accuracy.new_cases_MSE
1-10
 of 40
Config parameter
Importance
Correlation
model_params.seq_len
model_params.seq_len
batch_size
batch_size
dataset_params.forecast_history
dataset_params.forecast_history
training_params.batch_size
training_params.batch_size
dataset_params.batch_size
dataset_params.batch_size
model_params.output_seq_len
model_params.output_seq_len
inference_params.dataset_params.forecast_history
inference_params.dataset_params.forecast_history
forecast_history
forecast_history
lr
lr
out_seq_length
out_seq_length
Loading...
Multivariate TS sweep (baseline, only new cases as features) -- Italy__Lombardy__81
﻿
Italy--Lombardy with new cases + mobility data
﻿
﻿
﻿
Multivariate TS sweep Lombardy new cases + mobility)81
﻿
Italy-- Lombardy with mobility + weather data
﻿
﻿
﻿
Run set81
﻿
Comparing the best run from each set (lombardy)
﻿
﻿
﻿
new cases + mobility + weather1
new cases1
new caes + mobility1
﻿
﻿

Add a comment