Utilizing Deep Learning to to aid in AML diagnosis and research
Deep learning is used to classify types of white blood cells in blood smear slide images to aid in AML diagnosis and research. We have built a tool to aid in Acute Myeloid Leukemia diagnosis because it is relatively less researched and early detection can save a lot of people’s lives. Specifically, our models identify white blood cells in single cell images with high accuracy. This aids in diagnosis as certain cell subtypes are indicative of disease.
Classifying Tweets Using CNN, BERT, and DistilBERT
Internet bot detection is an important challenge Twitter faces, and currently there is no policy banning Twitter bots. Using machine learning models and deep learning, Twitter should be able to flag bots on a tweet level. This project explores the use of BERT and DistilBERT models to classify if tweets were written by a bot or human. We also explore the use of LIME to identify which words infer a tweet came from an internet bot. This project’s results show great promise in the use of deep learning models to detect internet bots as well as recommendations on ways to adjust these models to detect tweets written by bots with higher accuracy.
Big Analytics Using Google BigQuery: Help Discover Business Insights for Lyft Bay Wheels
Sample pictures and SQL query to help discover business insights.I suggest either increasing the amount of the monthly or yearly membership fee or decreasing the amount of time per trip that is included as a part of the membership with no additional fee. I also suggest increasing marketing and awareness for the corporate memberships deal since most of the subsribers appear to be commuters to and from and work.
Hypothesis tests such as unpaired t-test, paired t-test, sign test, and Wilcoxon rank-sum test performed on various topics after exploratory data analysis such as in pictures above