search

choonghyunryu / dlookr

Tools for Data Diagnosis, Exploration, Transformation
https://choonghyunryu.github.io/dlookr/
209 stars 35 forks source link

Scan variables and detect problems early before running the eda reports #81

Open ashirwad opened 2 years ago

ashirwad commented 2 years ago

Hey @choonghyunryu,

I never had any difficulty running the diagnose_web_report(), but eda_web_report() and transformation_web_report() often give me problems! Unfortunately, the problems crop up after the report has already run for a few hours! To help isolate the problem, I run the reports with a subset of variables until I find out the variables that are causing problems. This process is very tedious! It would be nice if you could update the code for your eda report functions to first scan the variables to see if there are any variables that contain problematic values that may cause issues when compiling the report. If such variables are present, display an error message asking a user to first remove those variables and then compile the reports.

One error message that I often find is this:

Error in plot.window(...) : need finite 'xlim' values

Another error:

Quitting from lines 397-399 (eda_temp.Rmd)
Error in eval(expr, envir, enclos) : object 'factor_flag' not found

Thanks!

choonghyunryu commented 2 years ago

Hi @ashirwad,

Thanks for the very nice comment. It would be a pity if an error occurred after running for a long time.

After finding a problematic variable, in order to exclude it from the report or notify the user, it is necessary to define the problematic variable first.

What are the characteristics of the variable in question in the two error messages you presented?

Thanks!

choonghyunryu commented 2 years ago

Hi @ashirwad,

The second error you provided is a bug. (Error while performing eda_web_report() #83) I will solve this problem first.

choonghyunryu commented 2 years ago

Hi @ashirwad,

The following error was fixed in development version 0.6.1.9002.

Quitting from lines 397-399 (eda_temp.Rmd)
Error in eval(expr, envir, enclos) : object 'factor_flag' not found