Analysis of sales, purchasing, or production data using statistical and econometric methods to produce empirical and structured insights into markets and business operations.
Understanding business and market mechanics:
When trying to understand business mechanics, markets, and trade systems, it is essential to accurately capture and structure the relevant factors that drive outcomes.
I use rigorous empirical methods from data science and econometrics to structure dependencies. For example, I have used a range of regression estimators in the past, including Ordinary Least Squares (OLS), Various Logit Models, or Generalised Method of Moments (GMM) models, and I have employed techniques from time series econometrics (e.g. ARIMA-models, dynamic panel models, etc.) and data science (clustering with k-means or PCA).
Make accurate predictions:
For all models designed to produce accurate predictions of future demand, costs, and other business variables, a modern approach to forecasting and testing is the key to success.
I use best practice from big data analysis, such as automated out-of-sample testing (bootstrapping, decision trees), systematic tracking of performance variables (RSS, Probable Error), or Updating (discontinuity analysis, systematic retraining).
Identify key drivers of business success:
For models that are designed to identify the most important inputs to product design, product placement, or advertising, it is central to implement a robust and rigorous methods to identify causality.
I use methods that are robust to various forms of endogeneity (e.g. panel models, instrumental variables models, discontinuity analysis) and apply rigorous inference techniques (i.e. hypothesis testing).