參考資料 - 進階函數

和

NOTE
0 (零) 表示 False，其他值表示 True。
``````AND(logical_test1,[logical_test2],...)
``````

logical_test1

logical_test2

近似相異計數 (維度)

``````Approximate Count Distinct (dimension)
``````

比較計數函數

Approximate Count Distinct() 是改良 Count() 和 RowCount() 函數後的成果。其可將建立的量度用於任何維度報表，藉此演算不同維度項目的近似計數。例如，用於「行動裝置類型」報表中的客戶 ID 計數。

反餘弦 (列)

``````ACOS(metric)
``````

反正弦 (列)

``````ASIN(metric)
``````

反正切 (列)

``````ATAN(metric)
``````

指數迴歸：預計 Y (列)

``````ESTIMATE.EXP(metric_X, metric_Y)
``````

metric_X

metric_Y

Cdf-T

``````cdf_t( -∞, n ) = 0
cdf_t(  ∞, n ) = 1
cdf_t( 3, 5 ) ? 0.99865
cdf_t( -2, 7 ) ? 0.0227501
cdf_t( x, ∞ ) ? cdf_z( x )
``````

Cdf-Z

``````cdf_z( -∞ ) = 0
cdf_z( ∞ ) = 1
cdf_z( 0 ) = 0.5
cdf_z( 2 ) ? 0.97725
cdf_z( -3 ) ? 0.0013499
``````

上限 (列)

``````CEILING(metric)
``````

信賴度

``````fx Confidence (normalizing-container, success-metric, control, significance-threshold)
``````

餘弦 (列)

``````COS(metric)
``````

立方根

``````CBRT(metric)
``````

累積

``````| Date | Rev  | cumul(0,Rev) | cumul(2,Rev) |
|------+------+--------------+--------------|
| May  | \$500 | \$500         | \$500         |
| June | \$200 | \$700         | \$700         |
| July | \$400 | \$1100        | \$600         |
``````

累積平均值

NOTE

``````cumul(revenue)/cumul(person)
``````

指數迴歸_ 相關係數 (表格)

``````CORREL.EXP(metric_X, metric_Y)
``````

metric_X

metric_Y

指數迴歸：截距 (表格)

``````INTERCEPT.EXP(metric_X, metric_Y)
``````

metric_X

metric_Y

指數迴歸：斜率 (表格)

``````SLOPE.EXP(metric_X, metric_Y)
``````

metric_X

metric_Y

下限 (列)

``````FLOOR(metric)
``````

雙曲餘弦 (列)

``````COSH(metric)
``````

雙曲正弦 (列)

``````SINH(metric)
``````

雙曲正切 (列)

``````TANH(metric)
``````

IF (列)

``````IF(logical_test, [value_if_true], [value_if_false])
``````

logical_test

[value_if_true]

[value_if_false]

提升度

``````fx Lift (normalizing-container, success-metric, control)
``````

線性迴歸_ 相關係數

Y = a X + b。傳回相關係數

線性迴歸_ 截距

Y = a X + b。傳回 b。

指數迴歸_ 預計 Y

Y = a X + b。傳回 Y。

線性迴歸_ 斜率

Y = a X + b。傳回 a。

以 10 為底的對數 (列)

``````LOG10(metric)
``````

對數迴歸：相關係數 (表格)

``````CORREL.LOG(metric_X,metric_Y)
``````

metric_X

metric_Y

對數迴歸：截距 (表格)

``````INTERCEPT.LOG(metric_X, metric_Y)
``````

metric_X

metric_Y

對數迴歸：預計 Y (列)

``````ESTIMATE.LOG(metric_X, metric_Y)
``````

metric_X

metric_Y

對數迴歸：斜率 (表格)

``````SLOPE.LOG(metric_A, metric_B)
``````

metric_A

metric_B

自然對數

``````LN(metric)
``````

NOT

``````NOT(logical)
``````

logical

或 (列)

NOTE
0 (零) 表示 False，其他值表示 True。
``````OR(logical_test1,[logical_test2],...)
``````

logical_test1

logical_test2

Pi

``````PI()
``````

PI 函數沒有引數。

乘冪迴歸：相關係數 (表格)

``````CORREL.POWER(metric_X, metric_Y)
``````

metric_X

metric_Y

乘冪迴歸：截距 (表格)

`````` INTERCEPT.POWER(metric_X, metric_Y)
``````

metric_X

metric_Y

乘冪迴歸：預計 Y (列)

`````` ESTIMATE.POWER(metric_X, metric_Y)
``````

metric_X

metric_Y

乘冪迴歸：斜率 (表格)

``````SLOPE.POWER(metric_X, metric_Y)
``````

metric_X

metric_Y

二次迴歸：相關係數 (表格)

``````CORREL.QUADRATIC(metric_X, metric_Y)
``````

metric_X

metric_Y

二次迴歸：截距 (表格)

``````INTERCEPT.POWER(metric_X, metric_Y)
``````

metric_X

metric_Y

二次迴歸：預計 Y (列)

``````ESTIMATE.QUADRATIC(metric_A, metric_B)
``````

metric_A

metric_B

二次迴歸：斜率 (表格)

``````SLOPE.QUADRATIC(metric_X, metric_Y)
``````

metric_X

metric_Y

倒數迴歸：相關係數 (表格)

``````CORREL.RECIPROCAL(metric_X, metric_Y)
``````

metric_X

metric_Y

倒數迴歸：截距 (表格)

``````INTERCEPT.RECIPROCAL(metric_A, metric_B)
``````

metric_X

metric_Y

倒數迴歸：預計 Y (列)

``````ESTIMATE.RECIPROCAL(metric_X, metric_Y)
``````

metric_X

metric_Y

倒數迴歸：斜率 (表格)

``````SLOPE.RECIPROCAL(metric_X, metric_Y)
``````

metric_X

metric_Y

正弦 (列)

``````SIN(metric)
``````

T 分數

Z 分數的別名，即平均值偏差除以標準差

T 檢定

`X` 是 t 檢定的統計資料，且經常會是基於量度的公式 (例如 zscore)，並在每列進行評估。

1. 用其找出極端值：

code language-none
``````t_test( zscore(bouncerate), row-count-1, 2)
``````
2. 將其與 `if` 合併，以便忽略非常高或非常低的反彈率，然後統計其他項目上的造訪率：

code language-none
``````if ( t_test( z-score(bouncerate), row-count, 2) < 0.01, 0, visits )
``````

正切

``````TAN (metric)
``````

Z 分數 (列)

Z 分數的方程式為：

NOTE
μ (mu) 和 σ (sigma) 會自動從量度中計算得出。

Z 分數 (量度)

Z 檢定

NOTE

recommendation-more-help
080e5213-7aa2-40d6-9dba-18945e892f79