Menu ▾ ▴

#154 Excess and skewness calculated erraneous

v1.1.x
fixed
Erik Hänel
critical (96)
v1.1.0
Bug Fix
2017-02-11
2017-02-09
Erik Hänel
No

The skewness and the excess have to be divided by potences of the standard deviation to return the correct numbers. However the qualitative answer for the skewness is still correct, which might not be true for the excess.

Analysis:
Obviously the calculated statistical moments are not divided through the potences of the standard deviation, which is available in the variable dError. Add this division and check for the excess, when to substract the 3 from the so-called kurtosis.

Implementation:
The division through the potences of the standard deviation was implemented and the substraction of 3 was moved to the correct location.

Documentation:
Not needed - Bug fix.

Tests:
The stats of a normal distribution were calculated and the values for this distribution are now near zero as expected. Bug fix accepted.

Discussion

  • Erik Hänel

    Erik Hänel - 2017-02-09
    • status: open --> accepted
     

  • Erik Hänel

    Erik Hänel - 2017-02-09
    • Description has changed:

    Diff:

    --- old
+++ new
@@ -1 +1,9 @@
 The skewness and the excess have to be divided by potences of the standard deviation to return the correct numbers. However the qualitative answer for the skewness is still correct, which might not be true for the excess.
+
+**Analysis:**
+
+**Implementation:**
+
+**Documentation:**
+
+**Tests:**
    • status: accepted --> analyzing
     

  • Erik Hänel

    Erik Hänel - 2017-02-09
    • Description has changed:

    Diff:

    --- old
+++ new
@@ -1,6 +1,7 @@
 The skewness and the excess have to be divided by potences of the standard deviation to return the correct numbers. However the qualitative answer for the skewness is still correct, which might not be true for the excess.

 **Analysis:**
+Obviously the calculated statistical moments are not divided through the potences of the standard deviation, which is available in the variable `dError`. Add this division and check for the excess, when to substract the 3 from the so-called kurtosis.

 **Implementation:**
     

  • Erik Hänel

    Erik Hänel - 2017-02-09
    • labels: --> critical
    • status: analyzing --> implementing
     

  • Erik Hänel

    Erik Hänel - 2017-02-11
    • Description has changed:

    Diff:

    --- old
+++ new
@@ -4,7 +4,9 @@
 Obviously the calculated statistical moments are not divided through the potences of the standard deviation, which is available in the variable `dError`. Add this division and check for the excess, when to substract the 3 from the so-called kurtosis.

 **Implementation:**
+The division through the potences of the standard deviation was implemented and the substraction of 3 was moved to the correct location.

 **Documentation:**
+Not needed - Bug fix.

 **Tests:**
    • status: implementing --> testing
     

  • Erik Hänel

    Erik Hänel - 2017-02-11
    • Description has changed:

    Diff:

    --- old
+++ new
@@ -10,3 +10,4 @@
 Not needed - Bug fix.

 **Tests:**
+The stats of a normal distribution were calculated and the values for this distribution are now near zero as expected. Bug fix accepted.
    • status: testing --> fixed
     

Anonymous
Anonymous

Add attachments
Cancel





MongoDB Logo MongoDB