ConfidenceBand 置信带,置信区间,可信区间,基线,窗宽
曲线,折线上下多出来两条线,像个通道一样
置信带越窄,精度越高,但是可靠性越低
confidence interval 越宽说明包含真值的概率也就越大
confidence interval是对某个函数估计在某个特定的点而言的,
confidence band是对整个函数估计而言,
一种简单的构造confidence band的方法就是point-wise, 既将每一点的confidence interval简单叠加起来构成上下两条曲线;
所以 confidence band的解释和 confidence interval应该保持一致性,
就是对于同一个问题而言,选取的置信水平越大,confidence band覆盖的面积就会越宽
ConfidenceBand
echarts 置信带,基线 https://echarts.apache.org/examples/zh/editor.html?c=confidence-band
// 基准数据
const dataSource = [
{ label: '半导体', value: [] }
];
// 置信带数据
const source = [
{label: 'Boundary', value: [10,90]}
]
const confidenceBand = [
{
label: '边界',
value: source.map(it => it.value[0]),
// areaStyle: { origin: 'start', opacity: 0 },
lineStyle: { opacity: 0 }, // 开始的基线数据,隐藏,第二个数据在这上面叠加
stack: 'confidence-band',
symbol: 'none',
},
{
label: '边界',
value: source.map(it => it.value[1]),
areaStyle: { color: 'rgb(47,194,91)' },
lineStyle: { opacity: 0 },
stack: 'confidence-band',
symbol: 'none',
},
];
series数据
[
{
name: 'L',
type: 'line',
data: data.map(function (item) {
return item.l + base;
}),
lineStyle: {
opacity: 1
},
stack: 'confidence-band',
symbol: 'none'
},
{
name: 'U',
type: 'line',
data: data.map(function (item) {
return item.u - item.l;
}),
lineStyle: {
opacity: 0
},
areaStyle: {
color: '#eaf9ee'
},
stack: 'confidence-band',
symbol: 'none'
},
{
name: 'BaseLine',
type: 'line',
data: data.map(function (item) {
return item.value + base;
}),
itemStyle: {
color: '#333'
},
showSymbol: false
}
]