Simulation-Based Estimation and Prediction
3.3. Simulation-Based Estimation and Prediction#
Learning Outcome
Students will be able to use simulation methods to understand the implications of statistical models.
Sample Tasks
Estimate the probability of an event for a simple probability model from a simulation.
Assess the accuracy of an estimated probability.
Estimate the probability of an event in a complex statistical model from a simulation.
Estimate the conditional probability of one event given another event for models with a single component of variation and for models with more than one component of variation.
Estimate implications (features) of a statistical model. These features might be a percentile of a distribution for a specified value of the predictor, the regression coefficient itself, or the odds for logistic regression.
Make a prediction for new data from the model.
Our first readings, from Computational and Inferential Thinking [ADW21], continue from the readings in Section 3.1. They discuss the simulation of probabilities and conditional probabilities in a fun, real-world problem and then discuss computing and simulating probabilities more generally.
Reading Questions
Should you switch doors?
If you roll two 6-sided dice, what is the probability that their sum is even?
Our second reading, also from Computational and Inferential Thinking [ADW21], discusses how to use simulations to estimate percentiles.
Reading Questions
If 37 people take a test:
How many will score less than the median?
How many will score less than or equal to the median?
Our third reading, from Learning Data Science [LGN23], discusses simulating the probabilities in a classic model of drawing balls from an urn.
Reading Question
What is an urn?