Probabilistic projection of migration (setup)

We'll now introduce a demographic model to be used throughout the course. It projects net migration rates via an autoregressive model (AR1). It is a typical example of an embarrassingly parallel application where a set of estimated parameters is used repeatedly for prediction. In an AR1, a future value is dependent on the past value. The process has a long term mean (mu), a rate of convergence (phi) to mu, and an error term with standard deviation (sigma). You do not need to know details of the modelling.

We have simulated 1000 samples of these three parameters (dataset ar1est) and will use it to project the future distribution of migration rates. Starting from an initial rate (rate0), each row of ar1est can be used to generate a block of migration trajectories (using function ar1_block_of_trajectories()). This process is shown in the image below (where ar1_block_of_trajectories() is abbreviated with ar1_block()). Combining blocks from all rows gives the final migration distribution.

Your job in this exercise is to explore the function ar1_block_of_trajectories() (type ar1_block_of_trajectories in your console). Its argument id determines which row of ar1est to use. What type of object does the function return and what do the function arguments traj_len and block_size correspond to in the function output?

Possible answers

Matrix where argument traj_len corresponds to rows and block_size corresponds to columns.

Matrix where traj_len corresponds to columns and block_size corresponds to rows.

List of vectors where traj_len corresponds to list length and block_size corresponds to each vector length.

Exercise

Probabilistic projection of migration (setup)

Instructions

Possible answers