View Code? Open in Web Editor
NEW
Predicting First-Year Survival after Percutaneous Coronary Interventions: A Machine Learning-Based ShinyApp Web Application in R
Home Page: https://primace.aikadeh.com
License: Other
R 95.59%
Dockerfile 0.88%
JavaScript 0.85%
CSS 2.68%
primace's Issues
R-cmd-check2
should be removed,
check-test-coverage
should be corrected
badge address is not correctly determined
We can create a models archive dir in s3 bucket and put each trained model in it.
And another dir for better models.
it is necessary to add a scheduled web app availabity on daily basis
because the size of the docker image is large, and shinyapps.io host for free has limitted resources, so I suggest to add a simple webpage in shinyapps.io
host with a link to redirect users to primace.aikadeh.com
Add read_data_s3() function to utilities.R to read dataset object from aws s3 bucket
Applying load test on web site to assess its performance in high turn over situations
Footer part is ok in most resolutions, be in my mobile landscape it has extra unnecessary height.
To create a general dashboards for priviledged users who can are allowed to performs data entry
Time Dependent Score calculator as a reference for manual prediction of unknown target data
Github actions raise error when preprocessing()
triggers in the case of new datasets.possibly due to prepared dataset us provided to data_modification()
function
to ensure not pushing incorrect docker image create a separate docker-test
workflow for non-main branch
We have to replace Rscript{0}
with Rscript {0}
in install paws
and also last steps.
In prediction tool
, two boxes, prediction
table and performance
table do not have scrolling features. This resulted in overshoot of table in prediction table
. And potentially could affect performance
table
For prediction of individual we can add survival probability of each time also overal survival probabilty curv.
Webapp-availability workflow should detect non 200 response as success.
For step check-new-datase
of R-MLOps-check
workflow, we can use a shell script like as follows:
#! /bin/bash
echo " FILE_STAT=$( Rscript helper/check_lastModification.R) "
if [ $FILE_STAT == NULL ]
then
echo " no dataset detected"
elif [ $FILE_STAT == FALSE ]
then
echo " dataset has not been modified"
else
Rscript -e " primace::preprocessing()"
fi
Define correct path for writing binary file from s3 bucket to /root/tmp writable dir in liara docker
We can use official mariaDB
package to connect to database.
And use mariaDB
as DBaas
.
We can add a simple form in contact tab to connect to dbaas to collect users comments
At first admin should have a dashboard
It is better to put R-MLOps-Check
below the PRIMACE
webapp badge
a login page logo on the right upper corner should be added, with shinyauthr
package
It is better to put a badge for project instruction website built with pkgdown
to README file .
I think analytics page should have more informative content such as MACE number and MACE in each subgroub.
We need to have an automated workflow which check when main dataset has modified
it seems LIARA paas server has shut down, so it is necessary to change paas provider asap
We shoukd add a link to article in README
It is a good idea to put test coveage and license badge in readme.md
It is necessary to add a workflow with scheduled check for new dataset.
add Coronary to title, as percutaneous coronary interventions
It seems in readme file 7 ML model is written instead of 2 ML models.
For users page structure of website should be multipaged
It is necessary to add google analytics for
Current plot in analytics page has numerical vars as groups. It should be fixed
To hetter manage gh-pages a separate branch is better to be dedicated to it.
It is a good thing if a dark theme designed as main design or alternative to light design.
we can add github logo to link to gh repo, a google logo to link to gmail of [email protected] , link to artcile, etc...
Checking for new dataset workflow must be separated from main workflow.