Key Concepts: Clustering, Spectral Clustering, Linear + Nonlinear Dimensionality Reduction, Density Estimation, Gaussian Mixture Models / EM Algorithm, Naive Bayes, Logistic Regression, Support Vector Machines, Neural Networks, Feature Selection, Anomaly Detection, Random Forest, Adaboost, Bias Variance Tradeoff, Cross Validation
Favorite Topic: This was the course that truly made me understand PCA + ignite a new found interest in unsupervised learning methods. I absolutely loved the depth of this course. Every model presented was derived from scratch. Working through these rigorous proofs gave me immense satisfaction and made me nostalgic of my undergrad mathematics coursework.
Favorite Topic: Function Fitting. Learning about this topic was incredibly interesting since there are so many analytical methods which rely on optimization to find the parameters of best fit. After taking this course, I went back through some old textbooks and found myself much more equipped to understand the math behind these models.
Simulation
Key Concepts: Random variables, Probability Distributions, Random Number Generation, Monte Carlo Methods, Discrete Event Simulation, Poisson Processes, Markov Chains, Statistical Testing in Simulation Studies, Variance Reduction Techniques
Favorite Topic: Monte Carlo Methods. I found MC simulation to be an extremely powerful tool that is criminally underutilized. The best part about these models is just how simple they can be. Additionally, this topic helped me analyze data in a more distribution focused mindset instead of my previous reliance on point estimators.
Statistical Modeling and Regression Analysis
Key Concepts: Linear Regression, ANOVA, Logistic Regression, Ridge Regression, Lasso Regression, Elastic Net, Cross-Validation, Model Selection, Regularization, Interaction Terms, Residual Analysis, Variable Transformation, Feature Engineering, Model Diagnostics
Favorite Topic: Linear Regression. Regression models are synonymous with data analysis. This course was a great chance to revisit the fundamentals of regression that I learned in my undergraduate education. After returning to this topic years later, I found myself having a much higher appreciation for the beauty and depth of linear models.
Data and Visual Analytics
Key Concepts: Data Cleaning, Data Visualization, Classification, Clustering, Ensemble Methods, Spark, Text Analytics, Graph Analytics, Version Control, Data Integration, Scalable Computing
Favorite Topic: Engineering concepts in Analytics. Since joining Instacart, I've realized the impact that analytics can have when combined with the ability to deploy code. I am not a software engineer, but learning how to apply software engineering principles to data projects allows me to get an analysis out of a notebook and into a production environment. This course timed perfectly with my transition into the ads engineering team where I currently work.
Intro to Analytics Modeling
Key Concepts: Model Validation & Performance Measurement, Linear Regression, Generalized Linear Models, K Nearest Neighbors, Support Vector Machines, Decision Trees, Random Forest, Holt-Winters, CUSUM Anomaly Detection
Favorite Topic: Model Selection. Prior to this course, I had a hard time understanding and arguing why someone would choose one method over another. I believe this is a crucial skill that is hard to develop if you are learning various modeling methods separately. This course allowed me to combine areas of model performance, desired outcome, data structure, and domain knowledge in order to construct arguments justifying analytical techniques.
Intro to Computing for Data Analysis
Key Concepts: Python, Data Structures, Computational Complexity, Pandas, Numpy, Math to Code, Sparsity, Linear Regression (from scratch), K-means (from scratch), Logistic Regression (from scratch).
Favorite Topic: Math to Code. As someone who loves math, I could not get enough of taking a mathematical principle and building an algorithm to compute it. After completing this course I can confidently learn the intricacies of any analytical method by building it step by step.