Chapter Post-stratification as a tool for enhancing the predictive power of classification methods

It is well known that, in classification problems, the predictive capacity of any decision-making model decreases rapidly with increasing asymmetry of the target variable (Sonquist et al., 1973; Fielding 1977). In particular, in segmentation analysis with a categorical target variable, very poor imp...

Olles dieđut

Furkejuvvon:
Bibliográfalaš dieđut
Váldodahkkit: d'Ovidio, Francesco Domenico, D'Uggento, Angela Maria, mancarella, rossana, TOMA, Ernesto
Materiálatiipa: Online
Giella:eaŋgalasgiella
Almmustuhtton: Firenze University Press 2022
Fáttát:
Liŋkkat:ONIX_20220601_9788855184618_556
Fáddágilkorat: Lasit fáddágilkoriid
Eai fáddágilkorat, Lasit vuosttaš fáddágilkora!
_version_ 1869526473216360448
author d'Ovidio, Francesco Domenico
D'Uggento, Angela Maria
mancarella, rossana
TOMA, Ernesto
author_browse D'Uggento, Angela Maria
TOMA, Ernesto
d'Ovidio, Francesco Domenico
mancarella, rossana
author_facet d'Ovidio, Francesco Domenico
D'Uggento, Angela Maria
mancarella, rossana
TOMA, Ernesto
author_sort d'Ovidio, Francesco Domenico
collection Directory of Open Access Books
description It is well known that, in classification problems, the predictive capacity of any decision-making model decreases rapidly with increasing asymmetry of the target variable (Sonquist et al., 1973; Fielding 1977). In particular, in segmentation analysis with a categorical target variable, very poor improvements of purity are obtained when the least represented modality counts less than 1/4 of the cases of the most represented modality. The same problem arises with other (theoretically more exhaustive) techniques such as Artificial Neural Networks. Actually, the optimal situation for classification analyses is the maximum uncertainty, that is, equidistribution of the target variable. Some classification techniques are more robust, by using, for example, the less sensitive logit transformation of the target variable (Fabbris & Martini 2002); however, also the logit transformation is strongly affected by the distributive asymmetry of the target variable. In this paper, starting from the results of a direct survey in which the target (binary) variable was extremely asymmetrical (10% vs. 90%, or greater asymmetry), we noted that also the logit model with the most significant parameters had very reduced fitting measures and almost zero predictive power. To solve this predictive issue, we tested post-stratification techniques, artificially symmetrizing a training sample. In this way, a substantially increase of fitting and predictive capacity was achieved, both in the symmetrized sample and, above all, in the original sample. In conclusion of the paper, an application of the same technique to a dataset of very different nature and size is described, demonstrating that the method is stable even in the case of analysis executed with all data of a population.
format Online
id doab-20.500.12854ir-82026
institution Directory of Open Access Books
language eng
publishDate 2022
publishDateRange 2022
publishDateSort 2022
publisher Firenze University Press
publisherStr Firenze University Press
record_format ojs
spelling doab-20.500.12854ir-820262022-06-02T04:02:29Z Chapter Post-stratification as a tool for enhancing the predictive power of classification methods d'Ovidio, Francesco Domenico D'Uggento, Angela Maria mancarella, rossana TOMA, Ernesto Classification Asymmetry Post-stratification Predictive power It is well known that, in classification problems, the predictive capacity of any decision-making model decreases rapidly with increasing asymmetry of the target variable (Sonquist et al., 1973; Fielding 1977). In particular, in segmentation analysis with a categorical target variable, very poor improvements of purity are obtained when the least represented modality counts less than 1/4 of the cases of the most represented modality. The same problem arises with other (theoretically more exhaustive) techniques such as Artificial Neural Networks. Actually, the optimal situation for classification analyses is the maximum uncertainty, that is, equidistribution of the target variable. Some classification techniques are more robust, by using, for example, the less sensitive logit transformation of the target variable (Fabbris & Martini 2002); however, also the logit transformation is strongly affected by the distributive asymmetry of the target variable. In this paper, starting from the results of a direct survey in which the target (binary) variable was extremely asymmetrical (10% vs. 90%, or greater asymmetry), we noted that also the logit model with the most significant parameters had very reduced fitting measures and almost zero predictive power. To solve this predictive issue, we tested post-stratification techniques, artificially symmetrizing a training sample. In this way, a substantially increase of fitting and predictive capacity was achieved, both in the symmetrized sample and, above all, in the original sample. In conclusion of the paper, an application of the same technique to a dataset of very different nature and size is described, demonstrating that the method is stable even in the case of analysis executed with all data of a population. 2022-06-02T04:02:28Z 2022-06-02T04:02:28Z 2022-06-01T12:20:58Z 2021 chapter ONIX_20220601_9788855184618_556 2704-5846 https://library.oapen.org/handle/20.500.12657/56371 9788855184618 https://directory.doabooks.org/handle/20.500.12854/82026 eng Proceedings e report open access image/jpeg Attribution 4.0 International https://library.oapen.org/bitstream/20.500.12657/56371/1/26242.pdf Firenze University Press 10.36253/978-88-5518-461-8.24 10.36253/978-88-5518-461-8.24 2ec4474d-93b1-4cfa-b313-9c6019b51b1a 9788855184618 6 Florence open access
spellingShingle Classification
Asymmetry
Post-stratification
Predictive power
d'Ovidio, Francesco Domenico
D'Uggento, Angela Maria
mancarella, rossana
TOMA, Ernesto
Chapter Post-stratification as a tool for enhancing the predictive power of classification methods
title Chapter Post-stratification as a tool for enhancing the predictive power of classification methods
title_full Chapter Post-stratification as a tool for enhancing the predictive power of classification methods
title_fullStr Chapter Post-stratification as a tool for enhancing the predictive power of classification methods
title_full_unstemmed Chapter Post-stratification as a tool for enhancing the predictive power of classification methods
title_short Chapter Post-stratification as a tool for enhancing the predictive power of classification methods
title_sort chapter post stratification as a tool for enhancing the predictive power of classification methods
topic Classification
Asymmetry
Post-stratification
Predictive power
topic_facet Classification
Asymmetry
Post-stratification
Predictive power
url ONIX_20220601_9788855184618_556
work_keys_str_mv AT dovidiofrancescodomenico chapterpoststratificationasatoolforenhancingthepredictivepowerofclassificationmethods
AT duggentoangelamaria chapterpoststratificationasatoolforenhancingthepredictivepowerofclassificationmethods
AT mancarellarossana chapterpoststratificationasatoolforenhancingthepredictivepowerofclassificationmethods
AT tomaernesto chapterpoststratificationasatoolforenhancingthepredictivepowerofclassificationmethods