Feature generation for optimization of marketing campaign
Waqas, Muhammad (2020-06-23)
Waqas, Muhammad
M. Waqas
23.06.2020
© 2020 Muhammad Waqas. Tämä Kohde on tekijänoikeuden ja/tai lähioikeuksien suojaama. Voit käyttää Kohdetta käyttöösi sovellettavan tekijänoikeutta ja lähioikeuksia koskevan lainsäädännön sallimilla tavoilla. Muunlaista käyttöä varten tarvitset oikeudenhaltijoiden luvan.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:oulu-202006242653
https://urn.fi/URN:NBN:fi:oulu-202006242653
Tiivistelmä
Utilizing the gaming data for optimizing the entire gaming paradigm has revolutionized the thought process of developers and gamers alike. The significance of the gaming data can be judged from the fact that it is being used productively by the marketing agencies to develop algorithms that could predict the behavior of a certain gamer and the reaction to updates. The core idea behind the solution proposed and implemented in this thesis is focused on making the marketing campaigns more impactful. According to the facts from credible online resources, i.e., Statista.com, the business-to-business (B2B) organizations spent over $12.3 billion on marketing campaigns. Since one of the major aims of a marketing campaign is customer acquisition, which is also referred to as demand generation, measuring the success rate of the marketing campaign is also of great importance. Besides, the conventional Customer Relation Managers (CRMs) don’t have such features using which, the businesses can monitor the effectiveness of the marketing campaigns. The system this thesis proposes aims to analyze the gaming data, which can be used to extract features for refined marketing campaigns. To analyze and precisely classify the gaming data, this thesis proposes an algorithm running behind a full-fledged marketing campaign that can yield optimal results and which can be further refined to predict the future purchase behavior of the users in such marketing campaigns. To accomplish this task, the Random Forest Classifier is the one, which this thesis proposes and has been implemented to optimize feature selection in order to enhance the profit revenue of the business. The promising results of empirical research and studies have proven the capability of the random forest classifier, and after employing it in the research, it has been established that the mentioned classifier is absolutely capable of extracting significant features on the basis of the gaming data sets that were provided. More importantly, this study has indicated that the Random Forest classifier gives better results in predicting the purchase likelihood, which is an essential milestone for our project. It should be noted that the solution we have proposed does not only serve to predict the purchase likelihood, but it can also be preferably utilized for other aims and objectives which are related to optimizing the marketing campaigns.
Kokoelmat
- Avoin saatavuus [32009]