Kovenda Mbuale's Projects
To increase efficiency of a cotton mill. I set up an ANOVA 3 factor analysis model in R to determine best spindle & position that produces the longest roving. The only significant difference in roving length was observed when position was 3 and spindle was 1 or 2. (ANOVA Model in R)
I built an artist Tableau Dashboard, data pulled from youtube using Youtube API and Youtube Analytics API and stored in PostgreSQL database on AWS-RDS
Utilized Python's Object Oriented Programming Principles to build a cafeteria dinning credits system
To test if CAPTCHA secure login security codes are secure from bot attacks I developed a CNN classifier model in Python using TensorFlow to predict the CAPTCHA codes. Predicted over 40,000 CAPTCHA codes with an accuracy above 98%.
To decrease the time taken by a manual pre-OCR pdf labeling process. I built a CNN image classification model in Python using OpenCV, Tensorflow and Keras. The CNN model classifies PDFs according to the amount of noise, 1 - representing very high noise and 5 - representing clear pdfs.
To see if drivers were being profiled. I built a Support Vector Machine (SVM) classifier and a randomForest classifier to predict a driver's race given the traffic's stop's details. Successful classification will indicate the existence of bais in the traffic stops' data.
Config files for my GitHub profile.
R Shiny App to determine the factors that are most influential in patients’ survival of CHD. I created a Logistic Regression model in R using RStudio to predict the survival of CHD patients. Retrieved the data from the PHIS database using SQL & built tableau dashboards. The model predicted the survival of CHD with an AUC of over .90 and indicated that a higher EF % is a major factor.
To test if CAPTCHA secure login security codes are secure from bot attacks I developed a Random Forest classifier model in Python using Scikit-learn to predict the CAPTCHA codes. Predicted over 40,000 CAPTCHA codes with an accuracy above 97%.
A linear regression predictor with a primary goal of predicting the chance of admission for a candidate.