Recently Published
Mixed Media Model
Rundown of basic mixed media model using data from the datarium package.
MTA Modeling--Markov v Heuristic
Brief exploration and comparison of modeling multi-touch attribution across common digital channels. Markov chain methods and common heuristic or rules-based methods of conversion attribution are the focus.
Structural Vector AutoRegression Model: Inflation
This SVAR analysis looks at the relationship between the output gap, inflation, and interest rates in the United States. Of special interest here is the findings comparison using data from 1958 to 2022 and a subset of 1982 to 2022 which supports the importance of credible central banks.
Regression Discontinuity: Electoral Convergence Theory
This is a replication of work done by Lee, Moretti, and Butler from their 2004 piece entitled "Do Voters Affect or Elect Policies: Evidence from the US House". It uses RDD to exploit the 50% threshold of winning elections in an attempt to gauge whether convergence or divergence theory in political economies is supported.
Markov Cohort Model
This Markov Cohort Model uses just three health states to follow cohort across 40 cycles. This model incorporates a varying transition probability matrix as well as a varying payoff matrix.
Machine Learning Case Study: Churn with RF and GLM
Case study analyzing churn for a telecom company comparing results from Random Forest and Generalized Linear Model methods.
Food Price Forecasting Function (datacamp Project)
Project from datacamp creating a function for forecasting prices from food pricing data across a collection of files.
Logit: Heart Disease Risk Factors
Logit model predicting binary heart disease outcomes based on sex, age, and maximum heart rate.
Forecasting US Natural Gas Consumption
Overview of univariate time-series forecasting leveraging ARIMA, ETS, and Naive methodologies.
Random Forests: Modeling NYC Taxi Fares
Brief examination of using tree and randomForest packages for predictive modeling.
Subclassification: Titantic
Applying subclassifications techniques to Titanic survivor data to address possible confounders of survival rate.
Synthetic Difference-in-Differences: Prop 99
Basic exploration of new SDiD causal method
Uber: Geospatial and Data/Time Analysis
Basic analysis of data from 2014 NYC Uber dataset with plenty of data visualizations.
Twitter Sentiment Comparison
Brief exploration of sentiment analysis on Twitter data