In mabafaba/hypegrammaR: A grammar of hypothesis test driven analysis

knitr::opts_chunk$set(
  collapse = TRUE,
  #,comment = "#>",
  eval=F
  )

Logic

Analysis Flow

The flow of an analysis according to the data analysis guidelines:

The implementation in hypegrammaR:

Main steps

Any analysis with HypegrammaR follows the same structure:

• Load your input data
• decide your analysis parameters
  • what type of hypothesis do you have?
  • what are the the dependent and independent variables, and what are their data types?
• map from your main parameters to a distinct analysis case.
• map your raw inputs and analysis case to receive an analysis result
• map your result to a visualisation / table or other appropriate outputs
• map your outputs to a file

Example

Once (ever per machine):

remotes::install_github('ellieallien/hypegrammaR',build_opts = c())
remotes::install_github('mabafaba/surveyweights',build_opts = c())
remotes::install_github('mabafaba/koboquest',build_opts=c())

Once per Assessment

Load the hypegrammaR package

library(hypegrammaR)

Load your files

All input files are expected as csv files.

Each input we usually expect with an assessment has it's own function to load it. They check that the format is in line with what is expected, make sure they play well with each other and prepare the functionality they are used for.

First the data. A csv file with data in standard Kobo format.

assessment_data<-load_data(file = "../data/testdata.csv")

Then a sampling frame. A csv file with one column with strata names, one column with population numbers. The strata names must match exactly some values in the data. We must tell the loading function which column is what in the sampling frame.

sampling_frame<-load_samplingframe("../data/test_samplingframe.csv")

Finally the questionnaire, which depends on the question and the choices sheet as a csv.

questionnaire<-load_questionnaire(data = assessment_data,
                                             questions = "../data/test_questionnaire_questions.csv",
                                             choices = "../data/test_questionnaire_choices.csv",
                                  choices.label.column.to.use = "label::English"
                                             )

Map to weighting

weighting <- map_to_weighting( sampling.frame = sampling_frame,
                               data.stratum.column = "stratification",
                               sampling.frame.population.column = "population",
                               sampling.frame.stratum.column = "strata.names",
                               data = assessment_data)

For each Hypothesis

Map to the analysis case

You need to know:

• The hypothesis type. Currently implemented are:
• direct_reporting
• group_difference
• limit
• the names of the columns that are your dependent and independent variables

case <- map_to_case(hypothesis.type = "group_difference",
                    dependent.var.type = "numerical",
                    independent.var = "categorical")

case

Now put it all together and Map to the result:

result<-map_to_result(data = assessment_data,
              dependent.var = "number_simultaneous_unmet_need",
              independent.var =   "region",
              case = case,
              weighting = weighting)

See the results

The map_to_result function gives you a number of things:

• A log message
• Meta information: a log of your inputs, what analysis case was used etc.
• The summary statistics (e.g. averages, percentages, ...)
• The hypothesis test results
• A visualisation

The log message

First, a message telling you how it went:

# result$message

That's what we want to see. If something went wrong, it should tell you here what happened.

meta information

# result$parameters

As you can see, it remembers what your input parameters were. It also added a standardised name of the analysis case.

The summary statistic

# result$summary.statistic

In this case, "numbers" are averages, because the input variable was numerical. min and max is the corresponding confidence interval. dependent.var.value give the corresponding variable values if they are categorical (NA otherwise.) The summary statistic will always be organised with exactly these columns, no matter what analysis you did. This is so that if you add a new visualisations or ouput format, it will work for any output from this function.

The hypothesis test

Next, there's information on which (if any) hypothesis test was used and the p value:

result$hypothesis.test

You'll probably be most interested in the p-value and the type of test that was used.

Add labels

Map to usable outputs

Chart

chart<-map_to_visualisation(result)
heatmap<-map_to_visualisation_heatmap(result)
chart

For advanced users (that know ggplot): The visualisation function returns a ggplot object, so you can add/overwrite ggplot stuff; for example:

myvisualisation+coord_polar()

gimme Labels!

result %>% map_to_labeled(questionnaire) -> result_labeled
chart <- result_labeled %>% map_to_visualisation
heatmap <- result_labeled %>% map_to_visualisation_heatmap

Save as files

map_to_file(chart,"barchart.jpg")
map_to_file(result$summary.statistic,"summary_statistics.csv")

The grammar is built from two types of elements: - "Blocks": Take the output of a mapping - "Mappings": Decide what to do, call a "block" that does it, and returns another block.

mabafaba/hypegrammaR documentation built on Oct. 2, 2019, 11:33 a.m.