NaNs when using regressionPow
Opened this issue · 1 comments
hasnainali96 commented
I’m trying to run a power law regression using the data in the file attached
datapoints.txt
const d3 = require('d3-regression')
const regressionGenerator = d3.regressionPow()
.x(d => d.x)
.y(d => Number(d.y));
const output = regressionGenerator(data)
console.log(output)
which returns some NaNs, suggesting something is going wrong in the regression
[
[ 8, NaN ],
[ 26361, NaN ],
a: NaN,
b: NaN,
predict: [Function: fn],
rSquared: NaN
]
Is there a reason I'm seeing NaNs? What's the recommended way forward? Can we filter out certain data points to prevent this?
hasnainali96 commented
realised that this is because there are some y values which are 0 in the data, and NaNs are returned since ln(0)
is undefined.