Overview

Introduction

Beiwe-Analysis is a code repository for some of the existing Beiwe data analysis functions. Some of this code references Python or C-related code for speed, but most of this code is in R, and is readily converted into an R package. Note that this code here is distinct from the Beiwe data analysis platform, which runs this and other code in the background as scheduled by the user. More documentation to follow.

Table of Contents

• Overview
  • Preprocessing
  • Processing
  • Outputs
  • Utility

Overview

Preprocessing

Beiwe data comes in standard formats, but requires processing before generating any insights. This includes the most basic tasks, such as determining how much data one has, or the quality of the data. Also, it is common to have other data that one would like to combine with Beiwe data.

We say data is preprocessed if it takes the form of a two-dimensional array (i.e. matrix, data frame) with at least two columns:

• Datetime, a string containing a universal format for date and time. For now, it's a POSIXct object, but this may change.
• Person, a string containing an ID for the subject at hand.

The goal of preprocessing with raw Beiwe data or collaborator is to convert it to processed data. For now, all preprocessed data is small enough to be loaded into working memory.

Processing

Even after preprocessing, most analyses require basic steps before creating plots or performing statistical analysis. For example, plotting may require preprocessed data using several different datasteams, or statistical analysis may require comparing one time with a fixed time lag in the past. To avoid errors in these tasks, we list how we do them here. Data is called processed if it is preprocessed, and is the product of the functions specified here. The goal is to do as much processing as possible using only these functions.

Outputs

Once the data is processed, we arrive at the juicy part: plots and statistical analysis. These functions take processed data in a specified format as inputs, and return highly specific outputs such as plots, models, or tests. These functions should not contain any processing steps.

Utility

A great deal of functions call into the above categories, but some do not. All other helper functions belong in utility. Examples include R package dependencies, and global constants such as plotting colors.