Getting_Started

Getting Started

This small tutorial will give a general introduction to the package, show the installation and how a small simulation is performed.

1. Package Installation

The package can be installed through pip or conda - depending on the user's preference.

pip install adaptivetesting

or

conda install conda-forge::adaptivetesting

2. Load the items and set up the item pool

At first, the package should be imported. We recommend using a short alias, such as adt. Then, the item pool can be loaded from a compatible format, such as a dictionary or simple lists. In this example, we will load the item parameters from python lists.

import adaptivetesting as adt
item_pool = adt.ItemPool.load_from_list(
    a = [0.42, 0.3, 1.5],
    b = [0.5, 0.9, 1.1]
)

Here, we use a 2PL model which is a simplified version of the 4PL model so that the c and d parameters are not manually set but inferred by the package.

3. Define the adaptive test

The general adaptive testing procedure can be easily defined with the TestAssembler class. This class supports numerous arguments to also allow rather complex procedures. For this example, we will only focus on a very basic configuration.

We configure a test for a single participant using ML for the ability estimation and MFI for the item selection. Because we want to simulate the test later, we have to specify a true ability level (here 0) so that the package may draw corresponding response patterns. For reproducibility, we also set a seed (123).

test = adt.TestAssembler(
    item_pool = item_pool,
    simulation_id="test",
    participant_id="1",
    ability_estimator=adt.MLEstimator,
    true_ability_level=0,
    seed=123
    )

4. Set up the simulation

The simulation can now simply be set up with the Simulation class. Additionally, we have to specify a format in which the test results will be saved.

simulation = adt.Simulation(test, adt.ResultOutputFormat.CSV)

5. Run the simulation

To start the actual simulation, we have to specify a stopping criterion. We will use a standard error value of $0.4$ in this example.

simulation.simulate(criterion=adt.StoppingCriterion.SE, value=0.4)
simulation.save_test_results()