Skip to content
Comprehensive statistical analysis of data sets on the HP-41 calculator
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
src
DSTAT.mod
DSTAT.rom
README.md

README.md

hp-41_DSTAT.ROM (version 1B)

Comprehensive statistical analysis of data sets on the HP-41 calculator.

DSTAT is short for Data/STATistical analysis ROM.

HP-41: STATistical analysis galore

This ROM serves two programs; "DATA" for statistical data entry and management of both single and double variables, and "STAT" that does curve fitting and statistical analysis of single and double variable data sets. The "STAT" program has taken its main part from William Kolb’s book "Curve Fitting for Programmable Calculators" with the added basic statistcs, momentums, kurtosis and skewness of data. You will find Kolb’s excellent book over at hp-41.org.

DATA

This program simplifies the entry and handling of data sets, storing and retrieving data sets from extended memory (XM)

Enter the file name of the data set in Alpha and XEQ "DATA". If no file name is given or the file name given does not exist, you will have to XEQ C to create a new data set. A file name that includes the numbers 1 or 2 indicates whether the data set is for single values of Y or paired X & Y values and flag 1 or 2 will be set accordingly.

Upon execution, the program shows labels A-E and labels a-e as successive prompts:

+ 1:2 C:S >Σ V

- FX XG FM P

The meaning of each label:

Label: (prompt) Description
LBL A: + Adds data (single [X] or double [X&Y] depending on whether flag 1 or 2 is set) to XM file
LBL B: 1:2 Toggle single or double data entry (sets flag 1 or 2 respectively)
LBL C: C:S Makes file name in Alpha the current data set file or creates the file if file name is not found. Creates the file as a single or double variable set depending on whether flag 1 or 2 is set (remember to ensure the correct flag is set first – by using LBL B). In creating a data set, you must supply the first variable(s) as you would with LBL A to kick off the data set
LBL D: >Σ Calls the MCF+ routine in the “STAT” program (see below) and thereby processes the data set created for digestion by the “STAT” program, then jumps directly to the “STAT” program (see below). It start with the data point number in X and ends with the data point number in Y. So, if you want to analyze the points 5->16 in a data set, just do “16 ENTER 5” before pressing D. If you want to process the whole data set, ensure both register X and Y contains 0.
LBL E: V View the data set, with the Y-value in X and the X-value in Y. If flag 1 is set, the X-value shown is incremented from zero and upwards.
LBL a: – Remove the last entered data (like s-)
LBL b: FX Toggle flag 0. If flag 0 is set, it freezes the data set so that it becomes of fixed size. Newly entered data point(s) will then move the earliest data point(s) out of the data set. This is usefull for analysing moving averages, a set time period (such as quarterly or 12-months data), etc.
LBL c: XG XM Get; Get file (name in Alpha) from extended memory. After retreiving the file, new data points can be added (LBL A) to enlarge the data set (or if flag 0 is set, keep the size but move the earliest data out of the file)
LBL d: FM Go to the FILEMAN program
LBL e: P Shows Properties of the data set and Prints the data set on the attached printer as a graph. First it will show the Min/Max/median of Y and of X before it asks whether you want to commence printing. Press R/S to thus commence.
LBL I Shortcut to printing the data set as a graph. When using this shortcut, the Y-MIN and Y-MAX has not been calculated as with LBL e, so you have to enter it manually (Y-MIN [ENTER] Y-MAX)
LBL J Back to showing the manu for labels A-E and a-e

STAT

This program performs basic statistical analysis, moments, kurtosis, skewness and curve fitting (the 19 different curve types in Kolb’s book).

Upon execution, the program shows labels A-E and labels a-e as successive prompts:

+ >I BF Y BΣ

- .- DATA X *

The meaning of each label:

Label: (prompt) Description
LBL A: + Add one data point (like s+). This is for manual input of data points. Use the “DATA” program above if you want to work with data sets and have the ability to save and restore the sets to XM
LBL B: >I Enter the curve index (1-19) from Kolb’s book and get the function and its constants and correlation coefficient displayed. This will now be the selected curve type for showing X or Y (LBL D or d)
LBL C: BF Calculate the best curve fit and show the formula for that function along with its constants and correlation coefficient. This will now be the selected curve type for showing X or Y (LBL D or d)
LBL D: Y Calculate Y from the X given in the X register for the selected curve type
LBL E: BΣ Show the basic statistical data for the entered data set
LBL a: – Remove last data point entered (like Σ-)
LBL b: .- Remove the last data pair entered (like Σ-)
LBL c: DATA Goto the “DATA” program (see above)
LBL d: X Calculate X from the Y given in the X register for the selected curve type
LBL e: * Back to the main menu (prompts showing labels A-E and a-e)

RUNTST

Included in this ROM is a program from the User Library Solutions book, "Test Statistics" (called RUNTEST in that book). From the description (clarifications by me):

The Run Test for Randomness

Consider a sequence of symbols such that the symbols are of two types only. A run is a continuous string of identical symbols preceded and followed by the other symbol (or no symbol). For example, the sequence 1110100011 has five runs (111, 0, 1, 000 and 11).

Let the total number of runs in a sequence be "u", and let "n1" and "n2" represent the number of symbols of type 1 and type 2 respectively (n1 is 6 and n2 is 4 in the example above). If the sample sizes are large (say n1 and n2 are both greater than 10), then the randomness of the sequence can be tested using a z statistic which has the standard distribution. For small samples, the test is based on special tables. See the User Library Solutions book for equations and this link for more in-depth understanding: http://www.itl.nist.gov/div898/handbook/eda/section3/eda35d.htm

The program asks for the number of runs, the number of Type 1 and Type 2. It outputs "MU" (the mean) and "SIGMA" (the standard deviation) and Z (the test statistic). If the absolute value of Z is larger than 1.96 for larger sample sizes, the run is considered "not random".

This way of testing for randomity can be used to see if e.g. men and women are picked at random, if samples "above average" and "below average" are taken at random, etc. As long as you can break samples into two categories, this program can be used to test for randomity.

RG

Included in the ROM is also a random number generator. "RG" returns a random number between 0 and 1. This Random Number Generator uses TIME as input, rendering the numbers generated “more random” than the usual HP-41 solutions that keeps a seed stored for the next value generated. If there is no Time module, RG instead takes the value in X and Y as seeds.

License

This software is released into the Public Domain.

You can’t perform that action at this time.