A collection of projects authored by Aarthi, and likely shared out with the community as an open source project.
Created visualization using power BI and built dashboards to effectively tell a story about the number of active Facebook users and its trend.
Using a labelled data set (Wisconsin breast cancer diagnostic data) of cancer patients and the various attributes of a cell such as radius, texture, area, perimeter etc, a model was trained to predict the possibility of a cell being cancerous using Logistic Regression and K-fold cross validation.
Analysed the San Francisco crime database using the K-Means clustering algorithm, to predict the most crime-prone neighbourhoods.
Visualizing Infant mortality rate in countries across the world, by processing the data and plotting it on Google Earth.
Real-time surveillance video from traffic cameras was taken as an input, and was split into frames. An algorithm was designed to extract HOG and GLCM features from the images and classify using Multi-Class Support Vector Machines and Parallel Algorithms. This trained model was used to detect pedestrian-vehicle interaction.
Using a labelled data set comprising of Wikipedia’s Talk page comments, a model was trained to categorize toxic comments into two categories - toxic and threat. This model was then proposed as a means to auto-hide toxic and threat comments from social media platforms (chrome plug-in under development).
Performed statistical analysis of trending YouTube videos over time. Used numpy and pandas to pre-process the data, and after analysis used tableau to Visualize the results through a dashboard.