evaluatr.init: Initialize analysis

Description Usage Arguments Value

View source: R/analysis.R

Description

Initialize analysis

Usage

1
2
3
4
5
6
evaluatr.init(country, data, pre_period_start = "start",
  post_period_start, post_period_end = eval_period_end,
  eval_period_start, eval_period_end, n_seasons = 12,
  year_def = "cal_year", group_name, date_name, outcome_name,
  set.burnN = 5000, set.sampleN = 10000, denom_name,
  log.covars = TRUE, sparse_threshold = 5)

Arguments

country

A one-word label for output (eg country or state name).

data

Input dataframe. There should be a row for each time point (e.g., month) and category (e.g., age group). There should be variables for the date, for the category (or a column of 1s if only 1 category), for the outcome variable (a count), for the denominator (or a column of 1s if no denominator), and columns for each control variable.

pre_period_start

Date when analysis starts, YYYY-MM-01. defaults to first date in dataset.

post_period_start

Month when intervention introduced. YYY-MM-01

post_period_end

Date when analysis ends, YYYY-MM-01. defaults to first date in dataset. Defaults to end of evaluation period.

eval_period_start

First month of the period when the effect of intervention is evaluated. YYYY-MM-01. typically 12-24 months after post_period_start.

eval_period_end

Last month of the period when the effect of intervention is evaluated. YYYY-MM-01.

n_seasons

How many observations per year? Defaults to 12 (monthly data) Change to 4 for quarterly

year_def

Should results be aggregated by calendar year ('cal_year': the default) or epidemiological year ('epi_year'; July-June)

group_name

Name of the stratification variable (e.g., age group). If only one age group present, add a column of 1s to the dataset

date_name

Name of the variable with the date for the time series

outcome_name

Name of the outcome (y) variable in the 'data' dataframe. Should be a count

set.burnN

Number of MCMC iterations for burn in (default 5000),

set.sampleN

Number of MCMC iterations post-burn-in to use for inference (default 10000),

denom_name

Name of the denominator variable in the 'data' dataframe. if there is no denominator, include a column of 1s.

log.covars

Should the covariate be log transformed? (default: TRUE)

sparse_threshold

Threshold for filtering out control variables based on sparsity (mean number of cases per time period). Defaults to 5.

Value

Initialized analysis object, 'analysis' as described below

'analysis$country' as passed to 'country'

'analysis$input_data' as passed to 'data'

'analysis$n_seasons' as passed to 'n_seasons'

'analysis$year_def' as passed to 'year_def'

'analysis$pre_period' Range of dates in the pre-intervention period

'analysis$post_period' Range of dates in the post-intervention period

'analysis$eval_period' Range of dates in the evaluation period

'analysis$start_date' First date of the pre-intervention period

'analysis$intervention_date' Last time point before the start of the post-period

'analysis$end_date' Last date in the evaluation period

'analysis$group_name' as passed to 'group_name'

'analysis$date_name' as passed in in 'date_name'

'analysis$outcome_name' as passed in in 'outcome_name'

'analysis$denom_name' as passed in in 'denom_name'

'analysis$time_points' Vector of time points in the dataset

'analysis$set.burnN' as passed in in 'set.burnN'

'analysis$set.sampleN' as passed in in 'set.sampleN'

'analysis$log.covars' as passed in in 'log.covars'

'analysis$groups' Vector of groups analyzed

'analysis$sparse_groups' Vector indicating which groups were too sparse to analyze

'analysis$model_size' Average number of covariates included in the synthetic control model

'analysis$covars' Matrix of covariates used for analysis

'analysis$outcome' as passeed to 'outcome_name'

'analysis$sparse_threshold' as passed to 'sparse_threshold'


weinbergerlab/InterventionEvaluatR documentation built on Sept. 13, 2019, 3:51 p.m.