AttributeError: 'fast' object has no attribute 'get_data'

[monbiola]

Hi,

I am trying to make an Sensitivity analysis of a python model with FAST method.

My code is

... sampler = spotpy.algorithms.fast(spot_setup, dbname=PATH_DATA, dbformat='csv', parallel='mpi') sampler.sample(rep) results = sampler.get_data()

[I want to directly loaded results from the sampler instead of loading the results from the csv database because with this last method I have a memory error (file ~ 3GB).]

But I have the error

results = sampler.get_data() AttributeError: 'fast' object has no attribute 'get_data'

Is it a bug on my side or get_data method isn't implemented for fast Class ?

Thanks

[thouska]

Hi @monbiola, thank you for your message. The getdata function is implemented as a base function for all algorithms within spotpy. You can use it with:

results = sampler.getdata()

Just as a sidenote: Pandas gives you the possibility to load only parts of a file. E.g. if you are only interested to load the Objective function column from your file, you could use:

import pandas as pd results = pd.read_csv(PATH_DATA+'.csv', usecols = ['like1'], low_memory = True)

This might prevent a memory error.

[monbiola]

Hi @thouska,

thank you for your answer. I had seen in the code the getdata function is implemented as a base function for all algorithms, but for Fast algorithm it doesn't seem to work. I reproduce the bug by executing the code in example in the pkg (tutorial_fast_hymod.py). I have the 1.6.1 Spotpy version

import spotpy from spotpy.examples.spot_setup_hymod_python import spot_setup

if name == "main": parallel ='seq'

Initialize the Hymod example

spot_setup = spot_setup()

#Select number of maximum repetitions
rep = 1345

#Start a sensitivity analysis
sampler = spotpy.algorithms.fast(spot_setup, dbname='FAST_hymod', dbformat='csv')
sampler.sample(rep)

# Load the results gained with the fast sampler, stored in FAST_hymod.csv
results = spotpy.analyser.load_csv_results('FAST_hymod')

# Example plot to show the sensitivity index of each parameter
spotpy.analyser.plot_fast_sensitivity(results, number_of_sensitiv_pars=3)

# Example to get the sensitivity index of each parameter    
SI = spotpy.analyser.get_sensitivity_of_fast(results)

## Add for test
results_sampler = sampler.get_data()

spotpy.analyser.plot_fast_sensitivit(results_sampler,number_of_sensitiv_pars=3)

The error from this code

results_sampler = sampler.get_data() AttributeError: 'fast' object has no attribute 'get_data

Do you also have this error?

thanks for the tip to prevent memory error, I will try that too.

[thouska]

Hi @monbiola, I think, if you change the line

results_sampler = sampler.get_data()

to

results_sampler = sampler.getdata()

it should work

[monbiola]

Yes it works like that :) I based on the doc https://spotpy.readthedocs.io/en/latest/Sensitivity_analysis_with_FAST/ where it says get_data instead of getdata

Thank you !

[thouska]

Ah, perfect thank you. I changed the documentation accordingly now.