Importing and Exporting Spectral Data in skspec¶

Environment Setup¶

Configure notebook style (see NBCONFIG.ipynb), add imports and paths. The %run magic used below requires IPython 2.0 or higher.

Importing a Spectra Object¶

Because pandas does such a great job of IO (see pandas IO tutorial), skspec uses pandas IO with as little modification possible. The idea is to read your data into a pandas DataFrame, and then construct a Spectra from this. IE:

pandas.io $\rightarrow$ pandas.DataFrame $\rightarrow$ skspec.Spectra

Where the DataFrame is NOT MultiIndexed.

The two specific cases where skspec Spectra can be constructed directly from data are:

• From CSV
• From JSON

The reason being that skspec implements custom behavior for these approaches. Therefore, let's start there.

From CSV¶

We can construct a Spectra directly from a .csv file using Spectra.from_csv(*args,**kwargs), where any argument to pandas.io.readcsv() is valid. (see read_csv API here). In addition, the user may pass the following arguments:

• header_datetime: String formatting if header is datetimes
• index_datetime: String formatting if index is datetimes
• spectrakwargs: Remaining keywords are passed to Spectra constructor.

For example, in the cell below I've copy pasted the first four rows and columns of a CSV file where the column header is datetimes and the index values are wavelengths, and the delimiter is \t.

We can read this directly into a Spectra, and will specify that the index unit should be nanometers, and that the header is datetimes. We will have to specify the format, which is in this case "%Y-%m-%d %H:%M:%S". We'll read in one of the CSV files packaged in skspec.data

Had we used Spectra.from_csv() rather than TimeSpectra.from_csv(), the difference is that TimeSpectra stores columns as a custom TimeIndex while Spectra would have defaulted to a pandas DatetimeIndex.

From JSON¶

Because skspec has custom datastructures, it needs to overload the JSON behavior of¶

Talk about the directory parser? IE Spectra.spec_from_dir¶

Exporting a Spectra¶

IO Multi-Spectra¶

• skspec@groups.google.com
• hugadams@gwmail.gwu.edu
• twitter: @hughesadam87
• Mark Reeves Biophysics Group

References:

• [1] : REF 1
• [2] : REF 2

Related:

• skspec: Exploratory Spectral Data Analysis
• pyparty: Image Analysis of Particles
• Lorena A. Barba (GWU Engineering)
• xray: extended arrays for scientific datasets

Notebook styling ideas:

• Louic's web blog
• Plotly
• Publication-ready the first time: Beautiful, reproducible plots with Matplotlib