/plotly-scientific-plots

plotly-scientific-tools is meant to augment the plotly and dash visualization libraries for python. It is designed to combine rapid and beautiful data visualizations along with statistical analysis for research scientists and data scientists.

Primary LanguageJupyter NotebookMIT LicenseMIT

plotly-scientific-plots

This python library is meant to augment the plotly and dash visualization libraries. It is designed to facilitate rapid and beautiful data visualizations for research scientists and data scientists.

Its advantages over naive plotly are:

  • One-line commands to make plots
  • Integrated scatistical testing above plots
  • Expanded plot types (such as confusion amtrices, ROC plots)
  • more 'Matlab-like' interface for those making the Matlab --> python transition
  • Easily make full multi-figure dashboards in a single line using Dash

Requirements and installation

Required packages:

  • numpy
  • scipy
  • plotly
  • colorlover
  • dash
  • dash_core_components
  • dash_html_components

To install, simply use pip install plotly-scientific-plots

NOTE: to get latest version install directly from git, which is more commonly updated. Use the following: pip install git+https://github.com/rsandler00/plotly-scientific-plots.git

To import use import plotly_scientific_plots as psp

Examples and Usage

Plotly's key strength is its ability to do interactive visualizations. For scientists, this allows novel ways of exploring data with mouse clicks and hovers. To see a full list of plotly-scientific-tools examples and their descriptions, go through the examples.ipynb in nbviewer by clicking here

Below, are a limited set of examples to give the feel of how psp works:

Sample plots

Two dataset histograms:
psp.plot2Hists(data_source_1, data_source_2, names=['Data 1','Data 2'],
            normHist=True, title='Comparison of 2 Data Sources',
            KS=True, MW=True, T=True)

Notice that the statistics box only appears when mouse hovers on the databar of the given color.

plot2Hist_1

Scatter + Contour Plot:
psp.scatterHistoPlot(data_source_1, data_source_3, title='Contour of x_var & y_var', 
            xlbl='x_var label', ylbl='y_var label')

plot2Hist_1

Multiple Dataset Correlations + Stats:
psp.corrPlot([data_source_1, data_source_11, data_source_12], [data_source_3, data_source_31, 
            data_source_32], names=['datasetA', 'datasetB', 'datasetC'],addCorr=True, 
            addCorrLine=True, title='Correlation of x_var & y_var', xlbl='x_var label', 
            ylbl='y_var label')

plot2Hist_1

Polar Plot
psp.plotPolar([polar1], numbins=20, title='Polar Distribution')

plot2Hist_1

Dashboards

To make multi-figure dashboards simply collect all desired figures in a nested list. Each outer list will correspond to a column in the dashboard, and each figure within each outer list will be a row in the column. The pass that list to psp.startDashboard. A flask-based web-server will start showing the figures in the browser at the provided port (default port=8050). For example:

plot1 = psp.plotHist(..., plot=False)
plot2 = psp.plot2Hists(..., plot=False)
plot3 = psp.corrPlot(..., plot=False)
plot4 = psp.plotPolar(..., plot=False)
dash_plots = [
            [plot1, plot2],
            [plot3, plot4]
        ]
psp.startDashboard(dash_plots, port=8052)

An example dashboard appears below:

plot2Hist_1