repurpose package


repurpose.img2ts module

repurpose.ts2img module

module for conversion of time series data to image data Created on Mon Apr 20 11:08:58 2015


class repurpose.ts2img.Ts2Img(tsreader, imgwriter, agg_func=None, ts_buffer=1000)[source]

Bases: object

Takes a time series dataset and converts it into an image dataset. A custom aggregate function should be given otherwise a daily mean will be used

  • tsreader (object) – object that implements a iter_ts method which iterates over pandas time series and has a grid attribute that is a pytesmo BasicGrid or CellGrid
  • imgwriter (object) – writer object that implements a write_ts method that takes a list of grid point indices and a 2D array containing the time series data
  • agg_func (function) – function that takes a pandas DataFrame and returns an aggregated pandas DataFrame
  • ts_buffer (int) – how many time series to read before writing to disk, constrained by the working memory the process should use.

does the conversion from time series to images

tsbulk(gpis=None, **tsaggkw)[source]

iterator over gpi and time series arrays of size self.ts_buffer

  • gpis (iterable, optional) – if given these gpis will be used, can be practical if the gpis are managed by an external class e.g. for parallel processing
  • tsaggkw (dict) – Keywords to give to the time series aggregation function

  • gpi_array (numpy.array) – numpy array of gpis in this batch
  • ts_bulk (dict of numpy arrays) – for each variable one numpy array of shape (len(gpi_array), len(ts_aggregated))

repurpose.ts2img.agg_tsmonthly(ts, **kwargs)[source]
  • ts (pandas.DataFrame) – time series of a point
  • kwargs (dict) – any additional keyword arguments that are given to the ts2img object during initialization

ts_agg – aggregated time series, they all must have the same length otherwise it can not work each column of this DataFrame will be a layer in the image

Return type:


Module contents