Topics: |
The NORMSDST and NORMSINV functions perform calculations on a standard normal distribution curve. NORMSDST calculates the percentage of data values that are less than or equal to a normalized value; NORMSINV is the inverse of NORMSDST, calculates the normalized value that forms the upper boundary of a percentile in a standard normal distribution curve.
How to: |
Reference: |
The NORMSDST function performs calculations on a standard normal distribution curve, calculating the percentage of data values that are less than or equal to a normalized value. A normalized value is a point on the X-axis of a standard normal distribution curve in standard deviations from the mean. This is useful for determining percentiles in normally distributed data.
The NORMSINV function is the inverse of NORMSDST. For information about NORMSINV, see NORMSINV: Calculating Inverse Cumulative Normal Distribution.
The results of NORMSDST are returned as double-precision and are accurate to 6 significant digits.
A standard normal distribution curve is a normal distribution that has a mean of 0 and a standard deviation of 1. The total area under this curve is 1. A point on the X-axis of the standard normal distribution is called a normalized value. Assuming that your data is normally distributed, you can convert a data point to a normalized value to find the percentage of scores that are less than or equal to the raw score.
You can convert a value (raw score) from your normally distributed data to the equivalent normalized value (z-score) as follows:
z = (raw_score - mean)/standard_deviation
To convert from a z-score back to a raw score, use the following formula:
raw_score = z * standard_deviation + mean
The mean of data points xi, where i is from 1 to n is:
The standard deviation of data points xi, where i is from 1 to n is:
The following diagram illustrates the results of the NORMSDST and NORMSINV functions.
Many common measurements are normally distributed. A plot of normally distributed data values approximates a bell-shaped curve. The two measures required to describe any normal distribution are the mean and the standard deviation:
NORMSDST(value, 'D8');
where:
Is a normalized value.
Is the required format for the result. The value returned by the function is double-precision. You can assign it to a field with any valid numeric format.
NORMSDST calculates the Z value and finds its percentile:
DEFINE FILE GGPRODS
-* CONVERT SIZE FIELD TO DOUBLE PRECISION
X/D12.5 = SIZE;
END
TABLE FILE GGPRODS
SUM X NOPRINT CNT.X NOPRINT
-* CALCULATE MEAN AND STANDARD DEVIATION
COMPUTE NUM/D12.5 = CNT.X; NOPRINT
COMPUTE MEAN/D12.5 = AVE.X; NOPRINT
COMPUTE VARIANCE/D12.5 = ((NUM*ASQ.X) - (X*X/NUM))/(NUM-1); NOPRINT
COMPUTE STDEV/D12.5 = SQRT(VARIANCE); NOPRINT
PRINT SIZE X NOPRINT
-* COMPUTE NORMALIZED VALUES AND USE AS INPUT TO NORMSDST FUNCTION
COMPUTE Z/D12.5 = (X - MEAN)/STDEV;
COMPUTE NORMSD/D12.5 = NORMSDST(Z, 'D8');
BY PRODUCT_ID NOPRINT
END
The output is:
Size Z NORMSD ---- - ------ 16 -.07298 .47091 12 -.80273 .21106 12 -.80273 .21106 20 .65678 .74434 24 1.38654 .91721 20 .65678 .74434 24 1.38654 .91721 16 -.07298 .47091 12 -.80273 .21106 8 -1.53249 .06270
How to: |
The NORMSINV function performs calculations on a standard normal distribution curve, finding the normalized value that forms the upper boundary of a percentile in a standard normal distribution curve. This is the inverse of NORMSDST. For information about NORMSDST, see NORMSDST: Calculating Standard Cumulative Normal Distribution.
The results of NORMSINV are returned as double-precision and are accurate to 6 significant digits.
NORMSINV(value, 'D8');
where:
Is a number between 0 and 1 (which represents a percentile in a standard normal distribution).
Is the required format for the result. The value returned by the function is double-precision. You can assign it to a field with any valid numeric format.
NORMSDST finds the percentile for the Z field. NORMSINV then returns this percentile to a normalized value:
DEFINE FILE GGPRODS -* CONVERT SIZE FIELD TO DOUBLE PRECISION X/D12.5 = SIZE; END TABLE FILE GGPRODS SUM X NOPRINT CNT.X NOPRINT -* CALCULATE MEAN AND STANDARD DEVIATION COMPUTE NUM/D12.5 = CNT.X; NOPRINT COMPUTE MEAN/D12.5 = AVE.X; NOPRINT COMPUTE VARIANCE/D12.5 = ((NUM*ASQ.X) - (X*X/NUM))/(NUM-1); NOPRINT COMPUTE STDEV/D12.5 = SQRT(VARIANCE); NOPRINT PRINT SIZE X NOPRINT -* COMPUTE NORMALIZED VALUES AND USE AS INPUT TO NORMSDST FUNCTION -* THEN USE RETURNED VALUES AS INPUT TO NORMSINV FUNCTION -* AND CONVERT BACK TO DATA VALUES COMPUTE Z/D12.5 = (X - MEAN)/STDEV; COMPUTE NORMSD/D12.5 = NORMSDST(Z, 'D8'); COMPUTE NORMSI/D12.5 = NORMSINV(NORMSD, 'D8'); COMPUTE DSIZE/D12 = NORMSI * STDEV + MEAN; BY PRODUCT_ID NOPRINT END
The output shows that NORMSINV is the inverse of NORMSDST and returns the original values:
Size Z NORMSD NORMSI DSIZE ---- - ------ ------ ----- 16 -.07298 .47091 -.07298 16 12 -.80273 .21106 -.80273 12 12 -.80273 .21106 -.80273 12 20 .65678 .74434 .65678 20 24 1.38654 .91721 1.38654 24 20 .65678 .74434 .65678 20 24 1.38654 .91721 1.38654 24 16 -.07298 .47091 -.07298 16 12 -.80273 .21106 -.80273 12 8 -1.53249 .06270 -1.53249 8
WebFOCUS | |
Feedback |