py_isear_dataset's Introduction

py_isear_dataset

The International Survey on Emotion Antecedents and Reactions as a python dataset for MachineLearning http://www.affective-sciences.org/researchmaterial

Purpose

The purpose of this code repository is to provide a python loader for the Isear Data set The ISEAR dataset has been developped by the Swiss National Center of Competence in Research.

At first sight, it seems interesting as a text analysis dataset regarding emotions in language. Petty dabbler as I am currently, I'll let you be judge of it.

Basic documentation

the Isear.csv file

This file is the actual extract of the dataset, which was provided as an Access database. The data has been cleaned and normalized a bit

All value lists are including 0 as origin from now on, some results have been been downgraded by 1 to do so.

There are 3 undocumented (as far as I know) fields, which are Field1 Field2 and Field3 which I kept for practical purposes.

Isear fields and values : the enums.py file

Everything you should need to know regarding the labels and values for this dataset should be in the enums.py file

Usage

from py_isear.isear_loader import IsearLoader
attributes = ['EMOT','SIT']
target = ['TROPHO','TEMPER']
loader = IsearLoader(attributes, target, True)
data = loader.load_isear('path/to/Isear.csv')
data.get_data() # returns attributes
data.get_target() # returns target
data.get_freetext_content() # returns the text content of the database

py_isear_dataset's People