By leveraging US daily case count data, I apply Facebook's Prophet forecasting model to get predictions with a MAPE of 3% predicting case counts in the subsequent 3 day period which analyzes the impact of holidays on COVID-19 case counts - the findings may surprise you!

Utilizing a variety of ML techniques, including Neural Nets, GBM's, RF's, GAM's, and GLM's, I analyze the Boston Housing dataset to predict the median value of a home.

Customer conversion in a direct marketing campaign is one of the most important metrics of success when evaluating the campaign itself. In a world of limited resources, it’s often difficult to make the best use of a marketer’s time, we can improve on this by predicting whether a prospective customer will respond to a marketing campaign. In order to do this, I will: • understand which of the observed variables are most associated with the chance of subscribing to a term deposit (feature importance); • how the important variables relate to the predicted probability that a client will subscribe to a term deposit (feature effects); • build a model that could be useful in determining which future clients are likely to respond a term deposit, if contacted (prediction/deployment).

This modeling work predicts salary within 5k of actual salary reliably by utilizing MARS Modeling, Random Forests, and Tree Based Modeling approaches. Data source: https://www.kaggle.com/nikhilbhathi/data-scientist-salary-us-glassdoor

This is an overview of logistic regression techniques, covering the formation of the logistic regression optimization equation, assessment of variable importance to the model, automation techniques for model creation, and model comparison techniques.

Utilizing WDI (World Development Index) data, I go step-by-step through the creation of ARIMA models for time-series forecasting, and utilize machine learning to create an optimized model using the BIC & AIC. This optimally fits an ARIMA model to the data for GDP growth in France, the UK, and India.

Using 3 sets of information pertaining to dog adoptions, my team and I developed this analysis of the datasets, performing EDA, managing unnecessary/unusual values, analyzing a large set of mainly text data, and presenting it in useful forms through heat maps, word clouds, and other interesting tools.

This is the report I generated in my final project for my class on Statistical Methods - I analyzed a wine quality dataset, checked out distributions of data, and performed Bootstrap Aggregation to validate the Central Limit Theorem.

This is the assignment submission link for the 5th assignment of Data Wrangling on plotting in GGPlot. I used the txhousing dataset for this visualization and subsetted the data to only include the most recent year.

This model analyzes an alumni donations database by deploying Machine Learning with Linear Regression in R, resulting in a final R-Squared of 0.75, a 138% increase in the R-Squared vs. a standard Multiple Linear Regression Model of the data.

This markdown document comprises the submission document for the Data Wrangling mid-term evaluation for John Trygier, Yuqiu Wang, Dhruv Cairae, Carrie Vennefron, and Bob Koenig.

This document is an introduction to and description of my background and experience in analytics.