omair-a-khan / ddictr Goto Github PK

• Put Section titles on the headers of each page
• Put "Variable: " before variable. For example, "Variable: age"
• Put "Label: " before label. For example, "Label: Age at medhx.date, recalculated"
• Change "Derivation: " to "Description: "
• Put "Distribution: " before it plot, and move chunk_wrapper for plot right before summarize_data
• Provide Lower Unique Values and Highest Unique Values by epoch
• Update the last chunk title as "IDs of participants with missing values by Epoch: "

https://github.com/fullstackstatistician/ddictR/blob/61b64ddc47b1437806f6cb3598ccfd65ca3cecca/R/create_individual_entry.R#L1

https://github.com/fullstackstatistician/ddictR/blob/184213b9c255fc69c7a2d9c24908bd5a6ce1a8fd/variables.list.R#L101

Break frequency and percentage into two columns, right align the percentage and frequencies.

Because of a recent change to show plots by epoch, the method for calculating the density curve in the histogram incorrectly uses all the data to do this calculation rather than doing this separately for each epoch. This needs to be addressed so that each epoch has its own density curve.

https://github.com/fullstackstatistician/ddictR/blob/73060d32217d843f68fb8c424e112eaff4bc35bf/R/plot_distribution.R#L26-L44

Check whether the raw metadata has unit attribuite
If yes, add unit attribute to the merged rds data. For derived variables, we need to do it one by one.
Then include unit in the codebook.

https://github.com/fullstackstatistician/ddictR/blob/8cdb65d7b5acfb6ddd8093f1e8d412c00f8dd45a/variables.list.R#L76

https://github.com/fullstackstatistician/ddictR/blob/184213b9c255fc69c7a2d9c24908bd5a6ce1a8fd/variables.list.R#L112

Please update the column header from "p5" to "5%". The same applies to all percentiles.

https://github.com/fullstackstatistician/ddictR/blob/61b64ddc47b1437806f6cb3598ccfd65ca3cecca/R/summary_statistics.R#L8

In summarize_data, update numeric format using summary statistics across all epochs.

• Identify the smallest absolute value (denoted at v) in mean and sd, and the determine the format using log10(v).
• if v is >=10, i.e. log10(v)>=1, then no decimal places;
• if v is in [0.1, 10), i.e. log10(v) is in [-1, 1], then use format X.X;
• If v is in [0, 0.1), use the first significant digit after decimal which could be obtained using round(log10(v))+1. For example, np.executive.composite has mean of 0.007003 in Epoch 1. The first significant digit after decimal is the 3rd digit and round(log10(0.007003))+1=3

https://github.com/fullstackstatistician/ddictR/blob/61b64ddc47b1437806f6cb3598ccfd65ca3cecca/R/summarize_data.R#L20

No decimal places for percentage please.

https://github.com/fullstackstatistician/ddictR/blob/61b64ddc47b1437806f6cb3598ccfd65ca3cecca/R/summarize_levels.R#L19

https://github.com/fullstackstatistician/ddictR/blob/61b64ddc47b1437806f6cb3598ccfd65ca3cecca/R/plot_distribution.R#L20

Using option 1 in the link with count on the left Y-axis and density on the right Y-axis