## Welcome to AAPT Workshop W27 ### Machine Learning and PER *Supported by Michigan State University's College of Natural Science and The Lappan-Phillips Foundation.*

### Workshop Facilitators * Danny Caballero [caballero@pa.msu.edu](mailto:caballero@pa.msu.edu) * Rachel Henderson [hende473@msu.edu](mailto:hende473@msu.edu) [LML Tutorials Website](https://github.com/learningmachineslab/tutorials)

## Introductions * Name * Institution * Role * Experience with ML * Experience with Python

### What do you want to use Machine Learning for? [Google JamBoard](https://jamboard.google.com/d/18xCxRs7PLuf1Zbb0HClImXql8RHqVunD56YzBQ-BKbE/edit?usp=sharing)

### What do we use Machine Learning for? * Modeling assessment data * Finding structure in assessment data * Modeling student degree attainment * Determining factors that influence attainment * Modeling graduate admissions * Determining factors that predict admissions * Developing language models of papers * Extracting themes from a paper collection

### Today's workshop: * Short introduction to the ML paradigm * Guided tutorial to read, plot, analyze assessment data * Comparative analysis using ML regression methods

### Today's goals: 1. Develop a preliminary understanding of different ways to explore data using the ```pandas``` and ```matplotlib``` libraries 2. Become familiar with the ```model->train->score``` paradigm of ```sklearn``` 3. Develop a preliminary understanding of the code implementations of ```sklearn``` for regression models 4. Develop a preliminary understanding of how to evaluate these models.

### What does Machine Learning buy us (as education researchers)? * Data we have access to is increasingly complex, large, and fundamentally relational * Models from data science are able to handle a wider variety of data and predictions * Data science frameworks for model construction, development, validation, and comparison are robust * We can ask and answer new research questions *Caveats: ML is not always the right tool for the job
 Interpreting results from ML models can be a challenge*

### Exploring Data Tutorial We have collected assessment scores from students on a mechanics concept inventory. For this tutorial, we will guide you through reading in, plotting, and deciding on research questions. * [Link to Jupyter Notebook](https://github.com/learningmachineslab/tutorials/blob/master/exploring-data/data_exploration_tutorial.ipynb) * [Link to Data](https://github.com/learningmachineslab/tutorials/raw/master/exploring-data/data/regression_data.csv) * [Google JamBoard](https://jamboard.google.com/d/18xCxRs7PLuf1Zbb0HClImXql8RHqVunD56YzBQ-BKbE/edit?usp=sharing)

### Regression Tutorial With the same assessment scores, we will determine what factors contribute most in a model that predicts those scores. We will also evaluate different models to try to understand how robust our findings are. * [Link to Jupyter Notebook](https://github.com/learningmachineslab/tutorials/blob/master/regression/regression.ipynb) * [Link to Data](https://raw.githubusercontent.com/learningmachineslab/tutorials/master/regression/data/regression_data.csv) * [Google JamBoard](https://jamboard.google.com/d/18xCxRs7PLuf1Zbb0HClImXql8RHqVunD56YzBQ-BKbE/edit?usp=sharing) [Regression Methods in Python](https://scikit-learn.org/stable/supervised_learning.html#supervised-learning)

## Thank you! ### Contact information * Danny Caballero [caballero@pa.msu.edu](mailto:caballero@pa.msu.edu) * Rachel Henderson [hende473@msu.edu](mailto:hende473@msu.edu) [LML Tutorials Website](https://github.com/learningmachineslab/tutorials)