Thevesh

This account is verified. Learn more

@Thevesh

Mostly tweets data about Malaysia 🇲🇾
Curator of ElectionData.MY

Malaysia Joined August 2009

429 Following 23K Followers 13K Posts

Latest posts 5

Pinned post

Thevesh @Thevesh Jun 24

After almost 4 years of hard work, I'm very happy to share something which I hope will be a game-changer for election data in Malaysia:

electiondata.my

To cut a (very) long story short, the site gives you access to Malaysia's COMPLETE electoral history. Every election, every candidate, every party, every seat - all there at your fingertips.

There are 5 types of datasets available:
• Parlimen/DUN-level results (the 'official' results, covering every single election since 1955)
• Parlimen/DUN-level maps (covering every single boundary-drawing exercise since 1954)
• Anonymised individual-level voter rolls for GE-13, GE-14, and GE-15, as well as Johor 2022 and N9 2023 (more to come soon)
• Saluran (!!!!) level results for GE-13, GE-14, and GE-15, as well as Johor 2022 and N9 2023 (also more to come soon)
• DM-level maps for the latest set (versions going back to 2003 will follow eventually)

Enjoy!

Data-cleaning and curation is bitter work, but once it's done once, everyone can reap the benefits.

electiondata.my ElectionData.MY A public record of Malaysia's democracy since 1955. Independent, neutral, and freely accessible to all.

Thevesh @Thevesh 17h

And that's a wrap for voter turnout (VTO) tracking for today.

This was the first time I've run live estimates of VTO by ethnicity, made possible by @sprgovmy releasing realtime VTO by DM. Excellent work, and I hope repaid the better data with good output!

On the API side, ElectionData.MY has already served nearly 9 million hits, even before the results are out 😅 I expect the hammering to continue tonight, with 0 loss of performance!

As always, all pages + APIs + data lake files will be updated as official results are released.

If you want unofficial results, go follow your political party sites—maybe next election, SPR will release live data by saluran during counting?

Then I can estimate voter turnout AND results by age and ethnicity. What a dream that would be!

Bar chart titled "Live Voter Turnout (70.1% as of 18:00)", showing estimated turnout by ethnic group in the 2026 Melaka state election (SE-16, black) compared with the 2022 general election (GE-15, red).

- Malay (62.6% of voters): 78.1% in SE-16, vs 80.6% in GE-15
- Chinese (28.4% of voters): 58.8% in SE-16, vs 72.2% in GE-15
- Indian (6.5% of voters): 58.0% in SE-16, vs 77.5% in GE-15

The Indian bars are hatched to show these are the least certain estimates.

Turnout fell for every group compared with GE-15. Malay turnout barely moved (down 2.5 points), while Chinese turnout fell sharply (down 13.4 points).

Footnote: Estimated using ecological inference at polling district (DM) level, exploiting the differing geographic distribution of each ethnic group. 'Indian' is a smaller, more dispersed group, so its estimates are least certain.

Thevesh Verified account @Thevesh 21h

Quite a few people DMed me about this, so here's a quick 2-slider on how ElectionData.MY is generating live estimates of voter turnout (VTO) by ethnicity today.

TL;DR:
• There is large variation in ethnic distribution across DMs
• SPR is releasing VTO by DM
• This enables us to analyse the difference in VTO across DMs, based on ethnicity, and draw some conclusions (using a technique called ecological inference, pioneered in Malaysia by @dririshsea)

For this particular exercise, having data by saluran won't add that much value, because within a single Pusat Mengundi, there's not much difference in ethnic distribution across the diff salurans.

But if SPR had given us the saluran data, then I could have done live estimates of voter turnout by AGE, which is the other super crucial dimension.

@sprgovmy dah ambil langkah hebat, please just buat sampai habis! DM dah, next one saluran 🙏

cc @muhdsazlie, tq for asking!

View quoted post on X

Thevesh @Thevesh 21h

Quite a few people DMed me about this, so here's a quick 2-slider on how ElectionData.MY is generating live estimates of voter turnout (VTO) by ethnicity today.

TL;DR:
• There is large variation in ethnic distribution across DMs
• SPR is releasing VTO by DM
• This enables us to analyse the difference in VTO across DMs, based on ethnicity, and draw some conclusions (using a technique called ecological inference, pioneered in Malaysia by @dririshsea)

For this particular exercise, having data by saluran won't add that much value, because within a single Pusat Mengundi, there's not much difference in ethnic distribution across the diff salurans.

But if SPR had given us the saluran data, then I could have done live estimates of voter turnout by AGE, which is the other super crucial dimension.

@sprgovmy dah ambil langkah hebat, please just buat sampai habis! DM dah, next one saluran 🙏

cc @muhdsazlie, tq for asking!

A 2×2 panel of maps of Melaka, headed "Melaka's ethnic mix varies widely among DMs". Each map shades Melaka's 211 polling districts (DMs) by one ethnic group's share of registered voters, on a shared blue scale from 0% (lightest) to 100% (darkest), with DUN boundaries outlined. Malay (63% of voters statewide, majority in 148 DMs) is dark across most of the state, especially the north-west and the south-east coast. Chinese (28%, majority in 49 DMs) is concentrated in and around Melaka town on the central coast, with a few dark pockets inland. Indian (6%, majority in 2 DMs) is light in most places, with a few darker DMs in the north and east. Other (3%, majority in no DM) is near-white almost everywhere.

Source: electiondata.my voter roll for Melaka SE-16.

A 2×2 panel of scatter plots headed "Turnout rises with the Malay share of each DM", showing projected turnout as of 15:30 on polling day. In each panel, each dot is one of Melaka's 211 DMs, sized by number of voters. The x-axis is one ethnic group's share of the DM's voters (0–100%) and the y-axis is the DM's projected turnout. A fitted line runs through the dots and is extended as a dotted line to 100%, where a blue dot marks the group's estimated turnout. Malay: turnout rises from about 58% in DMs with no Malay voters to about 75% in all-Malay DMs; estimate 76.0%. Chinese: turnout falls as the Chinese share rises, from about 74% to 58%; estimate 59.4%. Indian: the dots cluster below 30% Indian share and the line slopes gently down; estimate 60.2%, slightly above the line's end. Other: the dots cluster below 15% and the line slopes steeply down; estimate 35.1%.

Methodology: Goodman ecological regression, raked to observed turnout in each DM.

Source: api.electiondata.my

Muhd Sazlie Ibrahim 🇲🇾 💉 @muhdsazlie 23h

@Thevesh Wah terbaik... Jangan lupa notify saya. 😅
View quoted post on X

Thevesh @Thevesh 1d

HUGE update:

For the first time ever, because of the speed + level at which SPR are releasing turnout data, I am able to estimate (using ecological inference) voter turnout by ethnicity every 5 minutes!

Not only that, but I'm also able to do an end-of-day projection by ethnicity as well. Latest projections:
• Malay: 72%
• Chinese: 66%
• Indian: 69%

Unfortunately cannot do by age, because SPR isn't releasing saluran-level data, but DM-level data is more than good enough to produce estimates by ethnic group.

All live and updated every 5 minutes here:

electiondata.my/elections/mlk/…

Bar chart titled "Live Voter Turnout (24.8% as of 10:35)", showing estimated turnout by ethnic group in the 2026 Melaka state election.

Each group has two bars: turnout now (10:35) and projected final turnout at 18:00.

Malay: 26.9% now, projected 71.9%
Chinese: 20.6% now, projected 65.6%
Indian: 23.3% now, projected 68.9%
Other: 21.5% now, projected 58.4%

Malay turnout is highest so far and Chinese the lowest.

A footnote says the figures are estimated using ecological inference at polling-district level, and that "Other" is a small, dispersed group, so its estimates are the least certain.

Thevesh Verified account @Thevesh 1d

It's Melaka election day!

For the first time ever, @sprgovmy is providing LIVE (literally realtime) voter turnout (VTO) via:

dashboard.spr.gov.my

However, as expected, no API or structured data for all of us to use and analyse.

Fixing that shortly, be right back.

If the data is what I think it is, this should also allow me to produce real-time estimates of voter turnout by ethnicity (not by age, because no VTO by saluran)!
View quoted post on X

Thevesh @Thevesh 1d

Live voter turnout chart is back on ElectionData.MY, and this time, the live figures + forecast are updated every 5 minutes ⚡️⚡️

Follow it here: electiondata.my/elections/mlk/…

Massive win from @sprgovmy making the data available.

Reminder that you can query the data via the Open API available here:
electiondata.my/openapi/endpoi…

Line chart of hourly voter turnout in Melaka from 08:00 to 18:00. As of 09:55, turnout in the 2026 state election (SE-16) is 18.5%. At 09:00 it was 9.9%, slightly ahead of the same hour in the 2022 general election (GE-15, 9.5%) and the 2021 state election (SE-15, 7.0%). A dashed forecast line projects SE-16 turnout reaching 71.5% by 18:00. That would be below GE-15's final 79.2% but above SE-15's final 65.9%.

Thevesh Verified account @Thevesh 1d

It's Melaka election day!

For the first time ever, @sprgovmy is providing LIVE (literally realtime) voter turnout (VTO) via:

dashboard.spr.gov.my

However, as expected, no API or structured data for all of us to use and analyse.

Fixing that shortly, be right back.

If the data is what I think it is, this should also allow me to produce real-time estimates of voter turnout by ethnicity (not by age, because no VTO by saluran)!
View quoted post on X