Building Statistical Models Using StatsModels

It is important to use statistical tools well to explore and analyze your data before using it in ML models. This course teaches how basic statistical models work and how you can use the StatsModel Python package for estimation and exploration.

It is becoming ever-easier to build and use machine learning models, but it still is a challenge to use those models intelligently, and without committing any egregious modeling errors. In this course, Building Statistical Models Using StatsModels, you will learn to intuitively understand how to approach statistical techniques and apply them without getting bogged down in arcane mathematics. First, you will delve into tests of statistical significance by using the T-test to see whether the differences in two samples of a population are different and how to tell if the differences are statistically significant. Next, you will explore how to use the Analysis of Variance (ANOVA) techniques to compare several different population samples to see whether they differ on the basis of single or multiple factors. Then, you will discover a number of different regression models, such as generalized (or weighted) least squares regression, which are typically used with heteroscedastic data and robust linear models to cope with outliers. Finally, you will learn specialized statistical models that work with time-series data, including autoregressive and moving average models, and the ARMA family which combines both of these. By the end of this course, you will have developed an intuitive understanding of statistics and will be able to apply that intuition to your own specific use case using the StatsModel Python library.