Category Archives: Data Science

Singapore Joe Looks for a House in Singapore. But it's too expensive

A SIBOR Forecaster: Can SG Joe Afford Housing Bank Loans?

I did up this SIBOR forecaster a couple of months as part of my John Hopkins Data Science Specialisation. The seed of the idea came from my own (very shallow) experience with comparing housing bank loans in Singapore.

I was trying to answer: “Can I still afford bank loans 10 years down the line?”

I felt like Singapore Joe…

Singapore Joe Looks for a House in Singapore. But it's too expensive

So I thought of forecasting SIBOR based on historical data.

I know I know. Historical Forecast Isn’t The Most Accurate

But it’s all I got.

So I scraped years of SIBOR data from MAS Domestic Interest Rates and  made a Shiny App in R.

SIBOR Forecaster at

Try it at

It’s a 1-month SIBOR Forecaster where you can choose:

  1. Length of Home Loans
  2. Start and End Dates for Forecasting Years

In addition to the graph, it also returns a table of possible SIBOR forecast tables.

Obviously, it needs to consider more features like FED rates and maybe the news. But it’s a start at predicting the future.

Now’s that a place where I’d like to live in.


Same Same but different.

Same Same But Different: How Different Are You From Your Partner? A Gallup Strengths Finder Visualisation

A’s friend recently explained my Gallup Strengths* results to me. It was insightful. But more importantly, he mapped my (E) strengths against my partner’s (A) strengths.

It turns out that we’re quite the opposite. She’s very strong in Relationship Building (people-oriented) while mine is a mix of Strategic Thinking and Execution (task-oriented).

I wondered: “To what degree are we different from each other?” 

So I built a Same Same But Different Visualisation.


I took the data (essentially a list of ordered pairs of strengths) and turned them into a scatterplot. The circles represent individual strengths. Their location determined by our order pairing. And the size represents the difference in our ordering.

As an example: for Discipline — E is 7, A is 32. The difference is 25. Hence it is located at 32 on A’s axis and 7 on E’s axis. It’s fairly large as it quite close to the maximum possible difference (34 -1 = 33).

Same Same but Different

For meaningful comparison, I drew a diagonal line (x=y) on the chart. This line shows how far we deviate from having the same strengths. As you can see, there are 6 strengths that are only 15% different (difference of 5) and only one of them — Intellection — overlaps in our mutual strengths. Interestingly, Relator falls just outside E’s strengths but within A’s strengths.

Same Same but Different Strengths Visualisation

What Does This Same Same But Different Viz Mean?

It highlights the degree of difference for each of our strengths. This shows where we complement each other and where we are lacking.

On the flipside, it also shows similarities that we can build on. In this case, it’s likely Intellection and Relator. The former supposes that we both enjoy thought-provoking debates and the later presumes trust and caring for each other.

At this point in time, this viz lays down “what is…”. I suspect that it can be tweaked to do a little more — what more? I can’t say off the bat. Still too new to Gallup Strengths  Finder.

You’re Both So Different. Will It Work?

There is great potential to complement each other. But at the same time there is also great potential for friction. We will see, decide and act through our preferred lenses. Many people have managed this friction with the right mix of tactics.

I’ve done this for myself.

Thinking on it, this visualisation can be used in almost any partnerships.  After all, being partners with someone is a little like marrying them.

Download the workbook from my Tableau Profile.  You’ll need Tableau 10 or greater to work on it though.

Appreciate any feedback and insights!

* The Gallup Strengths Finder measures your aptitude across 34 attributes. They are categorised into 4 areas: Executing, Influencing, Relationship Building and Strategic Thinking. More at 

Wikipedia Visualiser or a Galaxy of Stars? - a wikipedia visualiser
Like flying through a galaxy. Just that the planets are now Wikipedia entries.

Jaw-droppingly awesome visualiser of Wikipedia entries that imagines each entry as a planet. In fact, the lines (as you see from the screenshot) is a contextual link from one entry to another entry.

It’s pretty. More importantly, it’s a really decent attempt to “connect the dots” by joining nodes to contextual nodes.

View it at


“Where’s the Bus Leh?” Visualising How Singaporeans Ping SG BusLeh

It took a year (of me being mostly lazy) before I turned the really nice dataset from SG BusLeh [get it from iTunes | Google Play] into a visualisation.

I didn’t know what to ask. So I just explored the data willy-nilly. In part, to try Tableau on different datasets; in part, to quickly generate visualisations to find questions.

Here’s what I found:

  1. There’s a difference between how SG Bus Leh is used on weekdays and weekends.
    Most folks take it later on the weekends
  2. People seem to take a while to get “into the groove” after a weekend. On Mon and  Tue, folks ping the app during perceived rush hour  timings (7 to 830am). From Wed to Fri, they’re actually getting out of the house earlier
  3. Lotsa pings on Fri night. No surprises there

I’ve put up the visualisations on Tableau Public: How often is SG Bus Leh Used?

Appreciate any feedback on them!

Shouts to SG BusLeh (iTunes | Google Play) for releasing their data




Visualisations on Tableau Public: How often is SG Bus Leh Used?

When Possible, Visualise DATA

I’ve been prepping reports and cleaning datasets for the last few weeks.

It’s no different from my other data projects – except that I’m using Tableau to visualise the reports. I’ve always thought that visualising data was really all vanity.

Now I’m a convert. Visualising the data makes it so easy to see relationships between dimensions! The Viz below (go to Tableau Public for a live demo) is a Google Keyword Search & Competition Visualiser.


I built the Keyword Visualiser to answer this question:

I got 1,100 keywords from Google Keyword Planner.

How do I know which ones to focus my social listening efforts on first?

Continue reading When Possible, Visualise DATA

Visualising Data on Maps: Uses & Whys

For the last couple of weeks, I had created maps with  data overlays:

  1. Instagram map of Singapore General Election 2015
    Plots Instagram posts over electoral boundaries (See GE 2015 Instagram Posts Map)
  2. Cluster Map of Customers
    Plots & clusters individual customer locations to show concentration of customers in an area

I thought they were kind of cool. After all, I made those maps! Out of scraped data! With an automated tool (R, a programming language, is great for data scraping and stats work)!

And therein lied the problem: I fell in love with what I was doing without answering the all-important question

“So What?”

(from 3 Important Questions. Easy to give advice, harder to follow it)

I was stumped, indignant, and ashamed. I knew data on maps was useful but I couldn’t articulate it. Now with a cooler head, here’s a rundown of possible purposes behind maps with data overlays.

So What Can I Do With Data on a Map?

  1. Show Concentration & Spread
    The most obvious use – we can show how posts and Likes are clustered on top of an area of interest. Going along these lines, we can do the same for prices, traffic etc.


  2. Track Movement over Time
    Most geo-located posts have time-stamps. We could map a group of people’s postings over a time period. This could be used to identify user flow. Here’s a really creative visualisation of an NBA game on a map (CartoDB Blog: Displaying NBA Data).

  3. Show Before & After
    We can divide time-stamped data into categories to show differences between now and then. Like this spiffy map of travel time in the USA in 1880 and 1900 (CartoDB Blog: realtimeliness).


It occurred to me, as I’m writing this post,  that there’s actually more that we can do with data points on maps than just the usual cluster and spread.

What’s your use for data on maps?

Finding Dots: Getting and Making Sense of Data

Strategy needs information. Unfortunately, much of that information – particularly in Marketing – comprises of hand-me-downs or culled from a wish list. And yet, we live in an age that is choking with information, data, figures and statistics.

Thus it seems incongruous that strategic plans are so ill-informed in such an information age. That’s why I’ve taken up learning how to scrape and transform on- and off-line data into some kind of insight…some kind of evidence that sets a direction.

The Result

It hasn’t helped me with strategy work yet. But it is wow… just wow. I’ve learnt a new programming language (R), brushed up on my statistics, and made maps.

Actually, a map of Instagram posts and likes on the Singapore General Election.

So exciting!