Loading...
Paper Type
Complete
Abstract
Artificial Intelligence reveals great potential for enterprises e.g., intelligent services. However, small and medium enterprises struggle with Artificial Intelligence due to limited resources. Especially tasks such as survey response classification are yet not investigated. We address this research gap by means of a data science study. In particular, we analyze several baseline classification pipelines leveraging logistic regression, random forests, and linear support vector machines against wide headed CNN architectures with one-hot encoding or character embedding inputs. We find that the SVM model outperforms all other evaluated models in the setting at hand. In addition, we analyze the different predictions of the models and show typical prediction errors by means of a chord diagram of commonly misclassified brands.
Recommended Citation
Stein, Nikolai; Oberdorf, Felix; and Pirner, Jonas, "Convolutional Neural Networks for Survey Response Classification" (2020). AMCIS 2020 Proceedings. 39.
https://aisel.aisnet.org/amcis2020/data_science_analytics_for_decision_support/data_science_analytics_for_decision_support/39
Convolutional Neural Networks for Survey Response Classification
Artificial Intelligence reveals great potential for enterprises e.g., intelligent services. However, small and medium enterprises struggle with Artificial Intelligence due to limited resources. Especially tasks such as survey response classification are yet not investigated. We address this research gap by means of a data science study. In particular, we analyze several baseline classification pipelines leveraging logistic regression, random forests, and linear support vector machines against wide headed CNN architectures with one-hot encoding or character embedding inputs. We find that the SVM model outperforms all other evaluated models in the setting at hand. In addition, we analyze the different predictions of the models and show typical prediction errors by means of a chord diagram of commonly misclassified brands.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.