statismike / stenr Goto Github PK

As the R6 basics may not be the easiest for the newcomers, the vignette detailing the workflow is needed

Create a better scale menagement

• define a specific class to hold scale specification
• choose the scale from scale specification appended to CompFreqtable during usage of compute_scores method
• alternatively - support passing the FreqtableScale object into the CompFreqtable - automatically attach the scale
• create a dedicated method for new scale attachement

Create new methods or new class/subclass to calculate different norms for different groups (eg. different biological sex). Psychological questionnaires often have different norms for differing groups.

Tests for all methods

First testthat tests should be focused on running the code and getting all warnings for methods and functions:

• functions:
  • gen_freqtable function to call comp_freqtable$new() method and create new object of that class
• methods:
  • get_status
  • get_data_info
  • get_freqtables
  • compute_scores
  • get_scoretables
  • get_computed_scores
  • get_computed_scores_ext

Currently, both of these methods produce output in the form of a list of results. It is useful for passing the results programmatically, but not human-friendly.

It should produce similar human-friendly output as s3 methods for summary(), eg. summary.lm. It should be optional to get the current output for these methods too: get_status(as.list = T).

Currently, after initialization of the comp_freqtable object, it is impossible to modify the data kept inside. I think there is a possibility for the use case, when:

1. at the initialization, the initial base of raw scores are fed into the object
2. standard scores are being computed for the first participants to receive their results
3. after the first stage, the raw scores of the next participants could be fed into the object, scores recalculated with the new data, and the standard scores fetched from the object

So the new method is necessary which should be:

1. Possible only if there are already data kept
2. Appending new data to the existing private$source_data
3. Automatically recalculating freqtables, and calling self$compute_scores() for each standard score that was calculated before

As the current README is mainly a placeholder, it is needed to create new from scratch

Currently ScoreTable can be only created based on the collected raw data (using FrequencyTable) or simulated data based on the distribution description (SimFrequencyTable).

It seems to me as mandatory to also include some kind of slimmed up version - without the FrequencyTable object underlying its creation and usage

What will be gained:

• the ability to allow importing score table described by the authors of the questionnaire itself (very important!)
• the ability to calculate score table for different groups (eg. if in the columns will be values only for Male, Female, Other sex options, instead of other StandardScores
• the ability to export created ScoreTable to the format used in measures documentations

What will be lost:

• ability to recalculate ScoreTable to allow different StandardScores or appending new results
• appending ScoreTable to current implementation of CompScoreTable

Addition of GroupedCompScoreTable, mimicking the behaviour of CompScoreTable - with underlying GroupedFrequencyTable and GroupedScoreTable objects

Also exporting ScoringTable should be allowed

• CompScoreTable.R
• FrequencyTable.R
• GroupFuncs.R
• handle_raw_scores.R
• normalize.R
• ScoreTable.R
• ScoringTable.R
• StandardScale.R