Configuration Guide¶

DaSPi provides a centralized configuration system through the CONFIG object for managing global settings like language, username, and plotting styles.

Quick Start¶

Access configuration through the CONFIG object:

import daspi as dsp

# Set individual properties
dsp.CONFIG.language = 'de'
dsp.CONFIG.username = 'analyst'
dsp.CONFIG.style = 'ggplot'

Configuration Properties¶

Language¶

Controls the language for localized strings in charts and reports.

Supported languages: - 'en' — English (default) - 'de' — German - 'fr' — French

# Set language
dsp.CONFIG.language = 'de'

# Now strings are in German
print(dsp.STR.accepted)  # Output: 'akzeptiert'

Username¶

Sets the username displayed in chart annotations and info text.

# Set username
dsp.CONFIG.username = 'j4ggr'

# Username appears in chart info text
chart = dsp.SingleChart(
    source=df,
    target='measurement',
    feature='sample_id'
).plot(dsp.Scatter).label(info=True)

By default, the username is read from the USERNAME environment variable.

Style¶

Controls the matplotlib plotting style for all visualizations.

# Set plotting style
dsp.CONFIG.style = 'ggplot'

# All subsequent plots use ggplot style
chart = dsp.SingleChart(...).plot(...)

Available styles: - 'daspi' — DaSPi default style (recommended) - 'ggplot' — ggplot2-inspired style - 'seaborn-v0_8' — Seaborn style - Any matplotlib style name

Configure Multiple Settings¶

Use the configure() method to set multiple properties at once:

dsp.CONFIG.configure(
    language='fr',
    username='analyst',
    style='seaborn-v0_8'
)

Temporary Changes with Context Managers¶

Context managers allow temporary configuration changes that automatically revert when the context exits.

Temporary Language Change¶

# Default language is English
print(dsp.STR.accepted)  # 'accepted'

# Temporarily switch to German
with dsp.CONFIG.use_language('de'):
    print(dsp.STR.accepted)  # 'akzeptiert'
    # Generate German report here

# Automatically reverts to English
print(dsp.STR.accepted)  # 'accepted'

Temporary Style Change¶

# Default DaSPi style
chart1 = dsp.SingleChart(...).plot(...)

# Temporarily use seaborn style
with dsp.CONFIG.use_style('seaborn-v0_8'):
    chart2 = dsp.SingleChart(...).plot(...)
    # This chart uses seaborn style

# Back to DaSPi style
chart3 = dsp.SingleChart(...).plot(...)

Nested Context Managers¶

Context managers can be nested for complex workflows:

# Generate reports in multiple languages
for lang in ['en', 'de', 'fr']:
    with dsp.CONFIG.use_language(lang):
        # Generate report in current language
        model = dsp.LinearModel(...)
        dsp.ResidualsCharts(model).plot().label(info=True)

Reset to Defaults¶

Reset all configuration to default values:

dsp.CONFIG.reset()

This sets: - Language to 'en' - Username to environment variable or 'user' - Style to 'daspi'

Advanced Usage¶

Multilingual Reports¶

Generate the same analysis in multiple languages:

import daspi as dsp

df = dsp.load_dataset('painkillers-dissolution')
model = dsp.LinearModel(
    source=df,
    target='dissolution',
    factors=['employee', 'brand', 'catalyst']
)
model.recursive_elimination()

# Generate report in each language
for lang in ['en', 'de', 'fr']:
    with dsp.CONFIG.use_language(lang):
        chart = dsp.ResidualsCharts(model).plot().stripes().label(
            fig_title=f'Residuals Analysis ({lang.upper()})',
            info=True
        )
        chart.fig.savefig(f'residuals_{lang}.png', dpi=300)

Style Comparison¶

Compare different plotting styles:

styles = ['daspi', 'ggplot', 'seaborn-v0_8']

for style in styles:
    with dsp.CONFIG.use_style(style):
        chart = dsp.SingleChart(
            source=df,
            target='measurement',
            feature='sample'
        ).plot(dsp.Scatter).label(fig_title=f'Style: {style}')
        chart.fig.savefig(f'chart_{style}.png', dpi=300)

User-Specific Settings¶

Set user preferences at the start of a notebook:

import daspi as dsp

# User preferences
dsp.CONFIG.configure(
    language='de',
    username='reto.jaeggli',
    style='daspi'
)

# Now all analysis uses these settings
# ...