Hello,
The user is able to select an aggregation type in the frontend by clicking the down arrow on a number series (e.g. Sum, Min, Max, Count). I have identified that which method they pick is stored within expr.func (I hope this helps someone; it took me a while to find it). I built the following mapping to let me apply the correct calculation based on expr.func. But it won't work for stdev or variance. After some hunting, I realise that when Stdev or Variance is selected by the user, the values passed into options.dataViews[0].categorical.values[numberIndex].values are all zeroes (e.g. [0,0,0,0]). I would expect this if the numbers are identical, but they are not.
To put it another way, the following log console.log(options.dataViews[0].categorical.values[numberIndex].values); outputs [3000, 2000, 1000, 5000] if I have Sum selected, but [0,0,0,0] if I have Stdev selected. Is Stdev/variance not supported for custom visuals or is there some kind of rounding issue going on - and if so how do i fix it please?
export const AGGREGATION_METHODS = {
0: { name: 'sum', calculate: (values: number[]) => values.reduce((a, b) => a + b, 0) },
1: { name: 'avg', calculate: (values: number[]) => values.reduce((a, b) => a + b, 0) / values.length },
2: { name: 'count distinct', calculate: (values: number[]) => new Set(values).size },
3: { name: 'min', calculate: (values: number[]) => Math.min(...values) },
4: { name: 'max', calculate: (values: number[]) => Math.max(...values) },
5: { name: 'count', calculate: (values: number[]) => values.length },
6: { name: 'median', calculate: (values: number[]) => {
const sorted = values.slice().sort((a, b) => a - b);
const mid = Math.floor(sorted.length / 2);
return sorted.length % 2 !== 0 ? sorted[mid] : (sorted[mid - 1] + sorted[mid]) / 2;
}
},
7: { name: 'stdev', calculate: (values: number[]) => {
const mean = values.reduce((sum, v) => sum + v, 0) / values.length;
const variance = values.reduce((sum, v) => sum + Math.pow(v - mean, 2), 0) / (values.length - 1);
return Math.sqrt(variance);
}
},
8: { name: 'variance', calculate: (values: number[]) => {
const mean = values.reduce((a, b) => a + b, 0) / values.length;
return values.reduce((a, b) => a + Math.pow(b - mean, 2), 0) / values.length;
}
},
};
