- 👋 Hi, I’m @AndreasKarasenko
- 👀 I’m interested in Python, Data Science, Rock Climbing
- 🌱 I’m currently learning full stack development (mostly django), Recommendation Systems, SQL, Natural Language Processing (mostly transformers)
- 📫 How to reach me: [email protected]
Most of my publications are accompanied by their respective code repositories to ensure replicability.
| Title | Journal (VHB4 ranking) | Short description |
|---|---|---|
| Sentiment Analysis in Marketing – From Fundamentals to State-of-the-art | Marketing ZFP (C) | This paper presents a comprehensive sentiment analysis pipeline, validated through a systematic comparison of 12 machine learning models across diverse datasets, and demonstrates interpretability frameworks for understanding model classifications. Practical guidelines for model selection are provided, along with Python codes, to support researchers in applying the findings. |
| Measuring technology acceptance over time using transfer models based on online customer reviews | Journal of Retailing and Consumer Services (B) | This paper presents a novel approach to estimate the technology acceptance model (TAM) through online customer reviews, thus side-stepping the need for expensive surveys. We find that our machine learning model provides valid estimates of the TAM construct scores. This model can then be used to estimate the TAM for other products or services. |
| Beyond Sentiment Analysis: Comparing Models to Predict Technology Acceptance from Online Customer Reviews | Forthcoming in: Journal of Business Economics (B) | This paper extends the prior paper in two ways: 1. first we conduct a thorough empirical comparison to identify the ideal machine learning model for our case; 2. second we introduce a novel method: topic based TAM analysis. Topic based TAM analysis leverages BERTopic to produce comprehensive topic clusters alongside the estimated TAM construct scores to allow for fine-grained evaluation of topic specific issues. We find that our proposed method helps identify issues, that sentiment analysis alone would not be able to uncover. |