Blog on Rafael Widjajahakim

Hello World of Website Building

Thu, 01 Jan 2026 00:00:00 +0000

BACKGROUND STORY

Creating a portfolio website has been a personal project todolist for so many years. But, there were a lot of obstacles (mostly personal motivations) to clearly make one.

The more legit reasons are:

As many typical website builders (Wix, Squarespace, etc.) will require cost to host, which one is worth it to try for my very early stage build? Most of these are for eCommerce, and I don’t need many of these tools.
Is there a pre-made theme among these websites that fit to my need?
How to organize the website for what I can present/provide? I am not in Art & Design field where I have many pictures or showcase products.

A MORE ADVANCED PATH (why run before walk)

Sometime ago, I found portfolio website from someone in tech. The theme was not what I have seen in common website builders. At the footer of the page, it says ‘Built with Hugo, through Netlify". At that time, I have not coded with Python, Linux, or anything. I was not even familiar with Github etc. I just want to use that theme.

So, I tried to follow any articles I could find. But I could not finish testing and got a job that was not that technical anyway. So I thought, I wouldn’t need it anyway; so I stopped.

NOW

Fast forward, I found out that Github can host a simple static website through the pages. So, I guess it is time to finally check that todolist… except,…

MORE HEADACHE

{::comment}comment {:/comment}

Visualizing abstract formula for research paper

Wed, 02 Dec 2020 00:00:00 +0000

Line

tl;dr

Basically, AREDS clinical trial grading data was found to be flip-floping for our analysis purposes. We defined a certain criteria for our research that is unique but could cause confusion due to the complexity of the formula or uncommon technique in this dataset.

To address this, I created a simple visualization to help explain the flip-flop issue, and the result using our novel code.

Another good thing about this visualization is that, the code to make this graph can easily & quickly be modified depending if reviewers/leaders want to see a different version.

The original visualization was created using R + ggplot2.

BACKGROUND STORY

AREDS 1 overview

AREDS was a multiyear longitudinal clinical trial on age-related macular degeneration, looking to see the effect of supplements on the progression of the disease. During the first AREDS trial, the grading system utilized 9 steps plus progression to the advanced AMD.

Many papers have been published and explained the trial design and results. So I won’t go into details here.

–

Research Objective

While many studies now used the simplified scale, we used the full scale to create a more granular analysis on the time-based progression of the disease.

Severity scale 1-8: was for the non-advanced AMD grade.
Severity scale 9: was for the advanced AMD grade.
modified severity scale 9, to include 10, 11, 12: was our modification to include GA and NV as part of the continuous scale.

–

Story

Since we analyzed it in very granular level, we hit an issue where the count of advanced AMD was different between timepoints.
Upon further investigation, we found that there were many cases where an eye was graded as progressed to advanced stage but then went back to non-advanced stage.
Since we wanted to analyze this progression like survival analysis, we had to make sure that a progressed eye should not go back to non-advanced stage.
- In some cases, the advanced grade only occured once in the middle of the study period and then back to very early stage (scale <6).
- In other cases, the time when the eye progressed to advanced stage occurred multiple times.
So, we developed a criteria to only select the eye as progressed eye, when the grade or status had passed the progression status/scale, at least at 2 time points in a sequence.
Then we realized the system was quite abstract.

So, I developed a visualization to show the progression and the flip flopping issue.

–

GRAPH PREVIEW

Understanding the graph:

The graph shows the typical grading for the subjects in AREDS 1 trial.
- The usual grade was taken after baseline visit was within 2 years.
- Different subjects had different total follow-up time.
The main issue is represented on ID 5, 6, and 7 (eye ID, not the same as the AREDS ID, the replacement is just for better visualization).
- Eye went to advanced stage on 2nd grading (scale 11), but went back to low (scale 4) on the 3rd visit. It even reached healthy status (scale 1) on the 5th visit.
- We decided that on these cases, the progression could be due to some human errors or study design error (blinded graders).
- These cases were the reason we developed a different criteria in defining when an eye was progressed to advanced stage.
The next few cases are representing how we selected the time of the progression.
- Eye 11 is a good example on how we skip the first progression (scale 11) as the following grades were back to non-advanced stage.
- When the eye finally had advanced stage grade back to back, we defined that this was a true progression.
- But instead of selecting the first grade, we chose the second timepoint for the true termination point.

Other version

CONCLUSION

Snippet for the R code is in my github.

The code originally was developed for our internal meeting where we wanted to view and compare individual ID/grades when we came across some issues. So, it was originally created so that we could quickly review each id and the code has more features such as multiple progression criteria.

We simplified the visualization a bit for the publication purpose.

CITATION

Seddon JM, Widjajahakim R, Rosner B. Rare and Common Genetic Variants, Smoking, and Body Mass Index: Progression and Earlier Age of Developing Advanced Age-Related Macular Degeneration. Invest Ophthalmol Vis Sci. 2020;61(14):32. doi:10.1167/iovs.61.14.32

Readme for Excel file

Sat, 22 Aug 2020 00:00:00 +0000

BACKGROUND STORY

While data analytic work now have plenty of platform and tools, many business are still using Excel as their main program/software/app.

Some of my works used Excel for a simple dashboard or even a work report. Sometimes I saw some Excel dashboard that is quite complicated and I have seen my colleagues were confused on how to use it.

In other cases, we received different datasets in Excel files. They could be a new data or just a transformed version of the larger data. But sometimes, we received the files almost every day due to the nature of the project. This has caused some confusions on what the data is about, or worse, the version of the revision (when the filename itself does not tell you about it).

SOLUTION

We tried to duplicate how a good data science project will have a README file. But for Excel, we can always have one tab for this readme, while the other tab is the content of the file.

Blog on Rafael Widjajahakim

Hello World of Website Building

BACKGROUND STORY

A MORE ADVANCED PATH (why run before walk)

NOW

MORE HEADACHE

Visualizing abstract formula for research paper

tl;dr

BACKGROUND STORY

AREDS 1 overview

Research Objective

Story

GRAPH PREVIEW

Understanding the graph:

The next few cases are representing how we selected the time of the progression.

Other version

CONCLUSION

CITATION

Readme for Excel file

BACKGROUND STORY

SOLUTION