SEUNG BEOM HAN
Indie Game Developer | South Korea
DATA SCIENCE PROJECTS
CRISP-DM using Python
A/B Testing Yonsei's AI Product
Analyzing Personalized Marketing Using CRISP-DM
I gathered some Google Analytics data regarding online advertisement conversion rates to conduct an A/B test and analyze its results to evaluate how effective personalized advertising is.
​
Multiple A/B tests were conducted as I segmented the dataset by marketing channels, age group, and language. Each A/B test consisted of dual conversion rate calculations (treatment and control), lift calculations, and statistical significance calculations.
​
The data was prepared and analyzed using Python on Google Colaboratory.
​
Testing the Effect of AlexNet
Text Classification and Analysis Using CRISP-DM
We gathered some textual data of numerous research papers published before and after the seminal AlexNet paper to assess how impactful AlexNet was on various fields of scientific research and publications.
​
The first part of the project was a classification task of creating if-then-else rules to classify papers into control and treatment. The TF-IDF vectorization algorithm was used for data preparation and the random forest model was used for classification training and predicting.
​
The second part of the project was a similarity and difference calculation task. Using the calculated cosine similarities, density curves were drawn, difference in differences technique was used to assess the differential effect of the AlexNet paper on other research papers.
​
The data was prepared and analyzed using Python on Google Colaboratory.
​