spotpy - A Statistical Parameter Optimization Tool

1.6.7 verified Sat May 09 auth: no python

spotpy is a Python library for statistical parameter optimization, calibration, and sensitivity analysis of environmental models. It provides a range of algorithms including Monte Carlo, Markov chain Monte Carlo (MCMC), Shuffled Complex Evolution (SCE-UA), DREAM, and various sensitivity analysis methods like FAST and Morris. Current version: 1.6.7. Release cadence is irregular, with several minor releases in recent years.

pip install spotpy

Common errors

error ModuleNotFoundError: No module named 'spotpy_setup' ↓

cause user tries to import spotpy_setup directly as a top-level module, but it is inside spotpy.examples.

fix

Use: from spotpy.examples.spotpy_setup_hymod import spotpy_setup as setup

error AttributeError: module 'spotpy' has no attribute 'algorithms' ↓

cause Old code that directly imports spotpy.algorithms without importing spotpy first, or incorrect import path.

fix

Import spotpy first, then use spotpy.algorithms.mc(...) or from spotpy.algorithms import mc

error ValueError: Unknown database format 'csv' ↓

cause Database format is case-sensitive. 'csv' is correct, but some users may accidentally use 'CSV' or 'Csv'.

fix

Use lowercase: dbformat='csv'

Warnings

breaking Migration to pyproject.toml in v1.6.1 removed setup.py. Installations via 'pip install -e .' may require setuptools update or use of 'pip install -e . --no-use-pep517'. ↓

fix Update pip and setuptools: pip install --upgrade pip setuptools. Or use 'pip install -e . --no-use-pep517' if needed.

deprecated Python 2 support was removed in v1.6.1. Python 3.6+ is required, but the current version specifies Python >=3.10. ↓

fix Upgrade to Python 3.10 or higher.

gotcha When using the DREAM algorithm, the 'random_state' parameter must be an integer for reproducible results. Passing None leads to non-reproducible chains. ↓

fix Set random_state to an integer, e.g., sampler = spotpy.algorithms.dream(..., random_state=42).

Imports

spotpy
```
import spotpy
```
The main module is imported directly as 'spotpy'.
spotpy_setup
wrong
```
import spotpy_setup
```
correct
```
from spotpy.examples import spotpy_setup as setup
```
spotpy_setup is a module inside spotpy.examples, not a top-level package.
demcz
wrong
```
import demcz
```
correct
```
from spotpy import demcz
```
demcz is a submodule inside spotpy, not a standalone package.

Quickstart

Basic Monte Carlo sampling using spotpy with a built-in example setup (HYMOD model).

import spotpy
from spotpy.examples.spotpy_setup_hymod import spotpy_setup as setup

# Create a test setup
spotpy_setup = setup()

# Select algorithm (here: Monte Carlo)
sampler = spotpy.algorithms.mc(spotpy_setup, dbname='MC_test', dbformat='csv')
sampler.sample(100)

# Get results
results = sampler.getdata()
print(results.keys())