Brand Preference Analysis using Unsupervised Language Models
Budget: ₹1,500 – ₹12,500 INR
You are expected to assume the role of a brand manager of any one firm and perform brand
preference analysis. For the analysis, extract relevant data of three firms. Treat one of them as a
focal firm and the other two as competing firms. Analyse customer behaviour patterns and present
your recommendations for the management of the focal firm. You are expected to use at least 250
reviews of each selected firm (not more than 1 year old review data). You may choose 3 different
firms or 3 different units of the same firm. e.g., Taj hotel reviews across 3 different cities or reviews
of 3 different hotels. For collecting recent review data, you are allowed to take external help (publicly
available crawler scripts, scripts from your group members or batchmates). The actual data analysis
scripts and report must be prepared based on your own original work.
Prepare a report – in not more than 12 pages, excluding title, references, and appendix – outlining
the steps you have followed while performing customer preference analysis. You can explore
different text analysis approaches (e.g. sentiment analysis, emotion analysis, cluster analysis) in
addition to topic modelling to understand customer behaviour and preferences. You should explain
your modeling, parameter tuning and model selection and evaluation steps clearly.
Your submission should include the actual data sets used for the analysis, data analysis report, along
with all the scripts/dashboards/programs used as part of the analysis.
preference analysis. For the analysis, extract relevant data of three firms. Treat one of them as a
focal firm and the other two as competing firms. Analyse customer behaviour patterns and present
your recommendations for the management of the focal firm. You are expected to use at least 250
reviews of each selected firm (not more than 1 year old review data). You may choose 3 different
firms or 3 different units of the same firm. e.g., Taj hotel reviews across 3 different cities or reviews
of 3 different hotels. For collecting recent review data, you are allowed to take external help (publicly
available crawler scripts, scripts from your group members or batchmates). The actual data analysis
scripts and report must be prepared based on your own original work.
Prepare a report – in not more than 12 pages, excluding title, references, and appendix – outlining
the steps you have followed while performing customer preference analysis. You can explore
different text analysis approaches (e.g. sentiment analysis, emotion analysis, cluster analysis) in
addition to topic modelling to understand customer behaviour and preferences. You should explain
your modeling, parameter tuning and model selection and evaluation steps clearly.
Your submission should include the actual data sets used for the analysis, data analysis report, along
with all the scripts/dashboards/programs used as part of the analysis.