datasets

Where datasets live.

This module gives access to several useful datasets. A dataset is represented as a class that inherits from base class Dataset. When instantiating such a class, you get an object with attributes:

• raw_data: data in the original file;
• data : data obtained after a pre-processing step was applied to the raw data.

The pre-processing step is performed by method preprocess of the class. For instance, for a regression dataset, the pre-processing steps normalises the predictors and adds an intercept. The pre-processing step of base class Dataset does nothing (raw_data and data point to the same object).

Here a quick example:

from particles import datasets as dts

dataset = dts.Pima()
help(dataset)  # basic info on dataset
help(dataset.preprocess)  # info on how data was pre-processed
data = dataset.data  # typically a numpy array

And here is a table of the available datasets; see the documentation of each sub-class for more details on the preprocessing step.

Dataset	parent class	typical use/model
Boston	RegressionDataset	regression
Eeg	BinaryRegDataset	logistic regression
GBP_vs_USD_9798	LogReturnsDataset	stochastic volatility
Nutria	Dataset	population ecology
Pima	BinaryRegDataset	logistic regression
Sonar	BinaryRegDataset	logistic regression
Neuro	Dataset	neuroscience ssm

See also utility function prepare_predictors, which prepares (rescales, adds an intercept) predictors/features for a regression or classification task.

Module summary

prepare_predictors	Rescale predictors and (optionally) add an intercept.
Dataset	Base class for datasets.
Nutria	Nutria dataset.
Neuro	Neuroscience experiment data from Temereanca et al (2008).
RegressionDataset	Regression dataset.
Boston	Boston house-price data of Harrison et al (1978).
BinaryRegDataset	Binary regression (classification) dataset.
Eeg	EEG dataset from UCI repository.
Pima	Pima Indians Diabetes.
Sonar	Sonar dataset from UCI repository.
LogReturnsDataset	Log returns dataset.
GBP_vs_USD_9798	GBP vs USD daily rates in 1997-98.