Skip to content

Visualization Tutorial

AeroViz provides rich visualization tools for aerosol data analysis and publication.

The matplotlib functions return (fig, ax); timeseries_interactive returns a Plotly figure. scatter, box, timeseries and diurnal_pattern take a time-indexed DataFrame; bar, violin and pie take pre-aggregated inputs (described with each function).

Basic Usage

from AeroViz import plot

# Scatter plot (x / y are column names)
plot.scatter(data, x='BC', y='PM25')

# Time series (y is the column, or list of columns, to plot)
plot.timeseries(data, y='BC')

Basic Charts

Scatter Plot

from AeroViz.plot import scatter

# Basic scatter plot
scatter(data, x='BC', y='PM25')

# With regression line
scatter(data, x='BC', y='PM25', regression=True)

# Color mapping
scatter(data, x='BC', y='PM25', c='RH', cmap='viridis')

Regression Analysis

from AeroViz.plot import linear_regression, multiple_linear_regression

# Linear regression (x, y are column names or lists of columns)
linear_regression(data, x='BC', y='PM25')

# Multiple linear regression (several predictors)
multiple_linear_regression(data, x=['BC', 'NO2', 'O3'], y='PM25')

Box Plot

box(df, x, y, x_bins=None) draws y grouped by x, choosing one of two modes automatically:

  • Categorical — x is non-numeric (a 'season' label, say) or x_bins is omitted: one box per unique value of x.
  • Binned — x is numeric and x_bins is given: x is cut at those edges (integer or float, any width, no rounding) and one box is drawn per bin.
import numpy as np
from AeroViz.plot import box

# Categorical: one box per season label
box(data, x='season', y='PM25')

# Binned: PM2.5 by wind-speed bins (0-2, 2-4, ... m/s)
box(data, x='WS', y='PM25', x_bins=np.arange(0, 11, 2))

# Binned with float edges
box(data, x='RH', y='PM25', x_bins=[0, 42.5, 65, 80, 100])

violin is the alternative when your data is already wide — one column per category (see below).

Bar Chart

bar(data_set, data_std, labels, unit, ...) — data_set is a DataFrame indexed by component name, with one column per group; data_std is the matching error DataFrame (or None).

import pandas as pd
from AeroViz.plot import bar

# Component contributions
components = ['AS', 'AN', 'OM', 'EC', 'Soil', 'SS']
data_set = pd.DataFrame({'PM2.5': data[components].mean()})  # index = components
bar(data_set, None, components, 'ug/m3')

Violin Plot

violin(df, unit, ...) — df is wide, with one column per category and each column holding that category's observations.

from AeroViz.plot import violin

# Distribution comparison across site types (one column each)
violin(data[['Urban', 'Suburban', 'Rural']], 'ng/m3')

Pie Chart

pie(data_set, labels, unit, style, ...) — data_set is a dict {group: values} (or a DataFrame), style is 'pie' or 'donut'.

from AeroViz.plot import pie

# Component proportions. The unit string is rendered as a mathtext label, so
# avoid a bare '%' (it fails to render) — use 'percent' or an escaped r'\%'.
pie({'PM2.5': data[components].mean().tolist()}, components, 'percent', 'donut')

Time Analysis Charts

Time Series

# Single variable
plot.timeseries(data, y='BC')

# Multiple variables on the primary axis
plot.timeseries(data, y=['BC', 'PM25', 'PM10'])

# Quick interactive Plotly view (one trace per column; toggle via legend)
plot.timeseries_interactive(data, columns=['BC', 'PM25'])

Diurnal Variation

# Single variable diurnal pattern (mean +/- spread by hour of day)
plot.diurnal_pattern(data, y='BC')

# Multiple variable comparison
plot.diurnal_pattern(data, y=['BC', 'PM25'])

Advanced Charts

Size-distribution heatmap

# Time × diameter heatmap of a dN/dlogDp matrix (index = time, columns = diameters)
plot.distribution.heatmap_tms(df_pnsd, unit='Number')   # 'Surface' | 'Volume' | 'Extinction'

Extinction contour

# Koschmieder-style fit: extinction as a power law of PM2.5 × gRH,
# drawn as a contour over the PM2.5 / gRH plane.
# df needs the scalar columns 'PM25', 'gRH' and 'Extinction'.
plot.contour(df[['PM25', 'gRH', 'Extinction']])

Wind Rose

# wind_rose lives in the meteorology submodule; WS / WD are column names
plot.meteorology.wind_rose(data, WS='WS', WD='WD')

# Color by a pollutant value
plot.meteorology.wind_rose(data, WS='WS', WD='WD', val='BC')

# Conditional bivariate probability function (pollutant by wind sector/speed)
plot.meteorology.CBPF(data, WS='WS', WD='WD', val='BC')

Correlation Matrix

# Correlation heatmap
cols = ['BC', 'PM25', 'PM10', 'NO2', 'O3']
plot.corr_matrix(data[cols])

Styling and saving

Drawing into your own axes, multi-panel layouts, fonts, colour-blind-safe palettes, journal column widths and savefig settings are collected in Publication Figures.