Accelarousal Study (NAT1) Methods

Getting Started

Prerequisites

• R (3.6.1)
• RStudio (1.2.5019)
• Required packages

Installing R Packages

To install required R packages, run script setup.R.

Folder Structure

• data: Raw and processed data.
• plots: Output plots of correlation matrices, clustering dendrogram, linear models, machine learning models.
• outputs: CSV/Text outputs.
• settings: Common setting of the project.
• scripts: Main script for analysis and data modeling.
• utils: Common ultility functions.

Configuration

All global settings are placed in settings/settings.R.

Analysis Timing Settings

The default time-series anlysis is using data of 30 seconds before to predict the class of next 5 seconds. You can change these values for different analysis.

TIME_PREV_SECONDS <- 30  # [5, 10, 15, 30]
TIME_NEXT_SECONDS <- 5

Plotly

As the project is using Plotly as a visualization tool, you can set up your Plotly account in settings/settings.R.

PLOTLY_USERNAME <- "<Your Plotly Username>"
PLOTLY_API_KEY <- "<Your Plotly API Key>"

Orca

The Plotly lib needs orca as an static image exporting backend. Follow instruction at https://github.com/plotly/orca for installation.

Script set

Correllation Analysis

• Run scripts/correlation.R

Clustering

• Run scripts/clustering.R

Linear Model

• Run scripts/linearModel.R

Additional Tools

Style Guide

http://jef.works/R-style-guide/

Install Styler

In RStudio, run following commands:

install.packages("styler")