postprocessing

Contains some helper functions for Marthe/PEST model postprocessing

Classes

PestPostProcessing

Helper class to manage some post-processing operation

Module Contents

class postprocessing.PestPostProcessing(pst)

Helper class to manage some post-processing operation (especially for plotting purpose).

phi_progress(pest_exe='++', log=True, phimlim=False, phimaccept=False, filename=None)

Plot the objective function (phi) evolution.

Parameters:

(str (filename) –
Can be:
- ’++’ : pestpp (-glm, -opt ,-swp, …)
- ’hp’ : pesthp
Default is ‘++’.
optional) (output file name to save plot on disk.) –
Can be:
- ’++’ : pestpp (-glm, -opt ,-swp, …)
- ’hp’ : pesthp
Default is ‘++’.
(bool (phimaccept) – Default is True.
optional) – Default is True.
(bool – Default is False.
optional) – Default is False.
(bool – Default is False.
optional) – Default is False.
(str – If None, the plot will not be saving on disk. Default is None.
optional) – If None, the plot will not be saving on disk. Default is None.
Returns –
-------- –
(AxeSubplot) (ax) –

Examples

ppp = PestPostProcessing(‘mypst.pst’) ppp.phi_progress(phimlim=True, filename=’phi_progress.png’)

phi_components(obs_groups=None, explode=0.05, filename=None)

Plot the contribution of each observation group to the total objective function.

Parameters:

(str/list[str] (obs_groups) – If None, all observation groups will be considered. Default is None.
optional) (output file name to save plot on disk.) – If None, all observation groups will be considered. Default is None.
(float (explode) – Default is 0.05.
optional) – Default is 0.05.
(str (filename) – If None, the plot will not be saving on disk. Default is None.
optional) – If None, the plot will not be saving on disk. Default is None.
Returns –
-------- –
(AxeSubplot) (ax) –

Examples

ppp = PestPostProcessing(‘mypst.pst’) ppp.phi_components(filename=’phi_components.png’)

export_stats(filename, cmap='YlGnBu', engine='excel')

Export/save some pest criterias (rss, mae, rmse, nrmse) for each observation groups as data table.

Parameters:

(str) (filename) –
(str (engine) – use on table values. Default is ‘YlGnBu’.
optional) (required exporting provider.) – use on table values. Default is ‘YlGnBu’.
(str –
Can be:
’excel’ (‘.xlsx’)

’latex’ (‘.tex’)

’html’ (‘.html’)
Default is ‘excel’.
optional) –
Can be:
’excel’ (‘.xlsx’)

’latex’ (‘.tex’)

’html’ (‘.html’)
Default is ‘excel’.
Returns –
-------- –

Examples

ppp = PestPostProcessing(‘mypst.pst’) ppp.export_stats(‘stats.xlsx’, cmap=’viridis’)

obs_vs_sim(kind='1to1', reifile=None, obs_groups=None, figsize=None, onefile=True, path=None)

Graphical analysis of observed vs simulated values.

Parameters:

(str (path) –
Can be :
- ’1to1’ : (x=measured, y=modelled)
- ’record’ : (x=observations, y=measured & modelled)
- ’residual’ : (x=observations, y=measured - modelled)
Default is ‘1to1’.
optional) (whatever export the produced plots.) –
Can be :
- ’1to1’ : (x=measured, y=modelled)
- ’record’ : (x=observations, y=measured & modelled)
- ’residual’ : (x=observations, y=measured - modelled)
Default is ‘1to1’.
(str – Format: ‘pst_name’ + ‘.rei’ + ‘n°itération’ (ex: ‘cal.rei5’) If None, the .res attriut in input pst wil be considered, generally it refers to the last iteration. Default is None.
optional) – Format: ‘pst_name’ + ‘.rei’ + ‘n°itération’ (ex: ‘cal.rei5’) If None, the .res attriut in input pst wil be considered, generally it refers to the last iteration. Default is None.
(str/list[str] (obs_groups) – If None, all observation groups with at least one not null observation will be considered. Default is None.
optional) – If None, all observation groups with at least one not null observation will be considered. Default is None.
(tup (figsize) –
Default is:
- ’1to1’ : (6,6)
- ’record’ : (9,5)
- ’residual’ : (9,4)
optional) –
Default is:
- ’1to1’ : (6,6)
- ’record’ : (9,5)
- ’residual’ : (9,4)
(bool (onefile) – Default is True.
optional) – Default is True.
(str –
If None, plots will not be saved. Otherwise:

If onefile is True: output file name to save pdf file on disk. If onefile is False: folder path to save plots on disk.

Default is None.
optional) –
If None, plots will not be saved. Otherwise:

If onefile is True: output file name to save pdf file on disk. If onefile is False: folder path to save plots on disk.

Default is None.
Returns –
-------- –
(dict) (axs) – Format: {loc_id: ax} (ex: {‘well_03’: <AxesSubplot>})

Examples

ppp = PestPostProcessing(‘mypst.pst’)

ppp.obs_vs_sim(kind=’record’, reifile=’cal_r_lizonne.rei9’,: obs_groups=[obg if obg.startswith(‘p’) for obg in ppp.nnz_obs_group], onefile=True, path=’obs_vs_sim_records.pdf’)
ppp.obs_vs_sim(kind=’1to1’, reifile=’cal_r_lizonne.rei13’,: onefile=False, path=’pest/post_proc’)

get_css_df(parnames=None)

Read Composite Scaled Sensitivities (CSS) as pandas DataFrame.

Parameters:

(str/list[str] (parnames) – If None, all adjustable parameters will be considred. Default is None.
optional) (required parameters full name(s).) – If None, all adjustable parameters will be considred. Default is None.
Returns –
-------- –
(DataFrame) (css_df) –

Format:

param1 .. paramN

iteration

1 7.733360e-08 .. 2.357000e-08 2 5.316330e-08 1.575150e-07 3 5.066230e-08 .. 5.618370e-09

Examples

ppp = PestPostProcessing(‘mypst.pst’) css_df = ppp.get_css_df()

export_css(filename, parnames=None, cmap='YlGnBu', engine='excel')

Export Composite Scaled Sensitivities (CSS).

Parameters:

(str) (filename) –
(str (engine) – use on table values. Default is ‘YlGnBu’.
optional) (required exporting provider.) – use on table values. Default is ‘YlGnBu’.
(str –
Can be:
’excel’ (‘.xlsx’)

’latex’ (‘.tex’)

’html’ (‘.html’)
Default is ‘excel’.
optional) –
Can be:
’excel’ (‘.xlsx’)

’latex’ (‘.tex’)

’html’ (‘.html’)
Default is ‘excel’.
Returns –
-------- –

Examples

ppp = PestPostProcessing(‘mypst.pst’) ppp.export_css(‘css.xlsx’)

static to_onefile(axs, filename)

Combine multiple plots into a unique pdf file.

Parameters:

(list) (axs) –
(str) (filename) – Note: must have the ‘.pdf’ extension.
Returns –
-------- –

Examples

df = pd.DataFrame(data={‘r’:np.linspace(5,58,50)}) axs = [df.plot(), (df*df).plot(), df.add(22).plot()] PestPostProcessing.to_onefile(axs, ‘all_plots.pdf’)

parameter_progress(parnames=None, pargroups=None, pest_exe='++', figsize=None, onefile=True, path=None)

Graphical analysis of the parameters values evolution along iteration from PEST calibration process.

Parameters:

(str/list[str] (pargroups) –

If None, all adjustable parameters will be considred.
Default is None.
optional) (whatever export the produced plots.) –

If None, all adjustable parameters will be considred.
Default is None.
(str/list[str] – If None, all adjustable parameter groups will be considred. Default is None.
optional) – If None, all adjustable parameter groups will be considred. Default is None.
(str (path) –
Can be:
- ’++’ : pestpp (-glm, -opt ,-swp, …)
- ’hp’ : pesthp
Default is ‘++’.
optional) –
Can be:
- ’++’ : pestpp (-glm, -opt ,-swp, …)
- ’hp’ : pesthp
Default is ‘++’.
(tup (figsize) – Default is (9,4).
optional) – Default is (9,4).
(bool (onefile) – Default is True.
optional) – Default is True.
(str –
If None, plots will not be saved. Otherwise:

If onefile is True: output file name to save pdf file on disk. If onefile is False: folder path to save plots on disk.

Default is None.
optional) –
If None, plots will not be saved. Otherwise:

If onefile is True: output file name to save pdf file on disk. If onefile is False: folder path to save plots on disk.

Default is None.
Returns –
-------- –
(dict) (axs) – Format: {parnme: ax} (ex: {‘hk_zpc_l11_z01’: <AxesSubplot>})

Examples

ppp = PestPostProcessing(‘mypst.pst’)

ppp.parameter_progress(pargroups=’soil’, onefile=True,: path=’soil_params_evolution.pdf’)
ppp.obs_vs_sim(parnames= [p if p.startswith(‘cap_sol’) for p in ppp.pst.adj_par_names()]: onefile=False, path=’pest/post_proc’)

__str__(): Internal string method.