The impact of campaign features on crowdfunding success in contemporary visual public art: a machine-learning application
Lazzaro, Elisabetta and Munim, Ziaul Haque and Nordgård, Daniel (2022) The impact of campaign features on crowdfunding success in contemporary visual public art: a machine-learning application. In: Tenth European Workshop on Applied Cultural Economics, 8-10 September 2022, University of Turin, Italy.
- Documents
- Details
This paper aims to shed light on the use of crowdfunding in the contemporary art market by studying the impact of main features of crowdfunding campaigns on their success for less tangible contemporary visual artistic expressions which might be more difficult to be traded through traditional art market channels, such as art public art. In order to disentangle possible benefits of crowdfunding for visual artists, we apply a machine-learning methodology to an ad-hoc taxonomy we elaborated, based on the general literature on crowdfunding benefits. An original dataset was built starting from Kickstarter category “Public Art” for the period 2009-2022. Using Data Robot, we estimated 36 ML models to predict raised funding amount in the public arts crowdfunding campaigns. Out of the estimated models, the Light Gradient Boosted Trees Regressor (Gamma Loss) (4 leaves) performed the best. Feature impact and feature fits were estimated after this model. Main results show that having a direct link of project realisation on the Kickstarter page, number of pledge levels, types of rewards, number of project updates, and number of backed projects by creator are top five features influencing the raised funding amount.
Actions (login required)
 |
Edit View |