Cody's Lab has shown how to extract potassium from bananas.

Spoiler: most of the potassium is in the peels.

https://yewtu.be/watch?v=fmaZdEq-Xzs

10 kg of bananas.

6.56g flesh, 3.44 kg peel.

6.56 kg banana flesh -> 1.5 kg dried bananas (plus about 100 g corn starch, used to prevent sticking)

Flesh pyrolised in a charcoal kiln to 0.5 kg banana charcoal. Peels -> 0.12 kg.

Charcoal burned down to 67g (flesh) and 51g (peels).

Potassium precipitated out as potassium chlorate. 33g flesh, 41g peels.

Decomposing out oxygen -> 36g potassium chloride (potash) (combined, flesh + peels). Radiation was measured at this point with a Geiger counter, at about double the background rate.

Metalic potassium produced through an alkalai converter, though only at about 50% yield: 9g produced rather than the 18g expected. The initial potassium chlorate yield was also below that suggested by FDA nutrient standards. Cody discusses this in the video.

Slow-mo exploding potassium banana sequence at end.

It was answered. You need 300 quadrillion bananas if you only use the energy emitted from the radioactive decay, or 10 bunches per day if you're willing to just burn them and basically ignore their radioactivity.

“Even if you captured that decay energy with perfect efficiency, powering a house would require about 300 quadrillion bananas, which would form a heap large enough to bury most of the skyscrapers in the NYC metro area”

300 quadrillion

The pile of bananas larger than Manhattan is how many you need to power a single house.

300 quadrillion, it's right in the text. Next to the visuals showing how large that pile would be.