Cost-Effective Deep Learning Approaches for Conversion Prediction in Social Media Marketing for Smes

https://doi.org/10.5281/zenodo.19121494

Abstract

In the world of marketing, small and medium-sized firms have a hard time making their social media campaigns work well when they do not have a lot of money to spend. This study looks at how some new deep learning methods can help these medium-sized firms predict when people will buy something from them and make the most of their marketing money. We used a set of data from digital marketing campaigns that we got from Kaggle and UCI which included things like what kind of campaign it was how big the company was, what kind of industry it was in how they advertised how much they spent, how people engaged with the campaign how many people bought something and if the campaign was successful. We tried out a few models to see which one could best predict if a campaign would be successful including some that used a lot of different methods together like Random Forest and Gradient Boosting and a simple neural network. We made sure the data was ready to use by making all the numbers similar and dealing with the problem of having of one kind of data than another. What we found out was that the models that used a lot of methods together were the best at predicting if people would buy something with Random Forest being right 94.37 percent of the time and Gradient Boosting being right 93.62 percent of the time. The neural networks were also very good at predicting especially when it came to remembering all the times when people did buy something, which shows they are good at finding patterns in marketing data. Our study shows that small and medium-sized firms can use learning and methods that combine a lot of different approaches to get a good idea of when people will buy something, which can help them make smart decisions about how to spend their marketing money and get a better return, on their investment. These results show just how much of a difference machine learning can make for companies that do not have a lot of money to spend on social media marketing.