Subscription Upgrade Prediction
Summary
The goal of this project was to try to utilize as many machine learning tools as possible to come up with the most accurate model for prediction. What the model was trying to predict was the users that would potentially switch from free to use software with ads to a paid subscription based on a couple of variables. This is beneficial to companies that want to help predict growth in the future or better understand the factors that lead to a user subscribing. The best performing model consisted of random forest, a 90/10 test/train split, and caret feature selection.
Tags: r, PYTHON, dyplyr, caret, xgboost, boruta, under-sampling, feature selection, backwards/forwards elimination, cross validation, Principal component analysis, data splitting, naive bayes, random forest, adaboost, Support vector, logistic regression, regression tree, ensemble.
Numeric Predictive Modeling
Summary
In this project I explore two different datasets to demonstrate real world applications of predictive machine learning. The first is data from customer spending and the second is data from emails to predict spam. numerous attributes and regression models were used to predict future customer spending and spam with an accuracy of roughly 92%.
Tags: R, Numeric Prediction, Maximizing Marketing, auc, logistic Regression, information gain, and feature selection.
Classification Modeling
Summary
In this project I analyzed 3 real world scenarios where classification would be utilized. The first is loan prediction and threshold adjustment. The second is maximizing profits from a marketing campaign. The third would be utilizing models to classify patients that have a malignant tumor.
Tags: R, Classification, predictive analytics, Maximizing Marketing, auc, Lift curve, roc curve, k-nn, and decision tree.
Production/Profit Optimization
Summary
For this Assignment, my partner and I had to optimize the cost of production of multiple products versus their profits using the language ampl. to do this, we had to use solvers like cplex that break down the code into linear algebra. Solvers can also be used in other languages like python.
Tags: AMPL, optimization, linear algebra, maximizing profits, solvers, and production analysis.
Chatbot
Summary
This is a chatbot that is themed around the comic book character ironman. The GUI was created in Python, and utilizes packages keras and tensorflow for Deep learning text capabilities. This chatbot also features built in features like background music, animations, and text to speech functions.
Tags: Python, GUI, chatbot, deep learning, Keras, Tensorflow, text-to-speech.