Statistical Thinking Notes fitted for exam

Description

Concise and detailed notes on all topics covered in end of semester exam, organised to help structure short answer and essay questions. It covers the following study points: - Different types of data can be collected, with corresponding visualisation methods and inferential procedures used depending on the situation - Having a simple random sample means we can apply a CLT and without knowing much about the shape of the population distribution -Frequentist inference treats data as random and parameters are fixed -Model fit checked using qq-plots, residual plots and other visualisations, R-squared, maximised log-likelihood, LOOCV -Forecasts typically condition on a single best estimate of theta and ignore uncertainty in this value -Bayesians treat parameters as random and observed data as being fixed -Summaries of relevant posterior distribution, minimise posterior expected loss -Having an sample means we can learn about the population shape