(01-16-2019 12:38 AM)zeno333 Wrote: [ -> ]The TI SR51 from 1976 could handle 99 data pairs...Geesh...and the 36X Pro just 42?...

I know what your topic is about, but show me somebody who need more than 10-15. Typically in engineering practice this number is 5-7-9. This is good enough to study something how depends on another property.

More important to sketching the measured data. I will very happy if one company will produce a good scientific calculator with stat plot possibility and data sharing to/from computer.

(01-21-2019 11:33 AM)Csaba Tizedes Wrote: [ -> ]I know what your topic is about, but show me somebody who need more than 10-15.

I remember one of my Professor fail any students 1 standard deviation below averages.

In other words, around 16% (1 in 6) will fail the class ...

I don't remember the class size, but it was definitely more than 15.

(01-21-2019 04:23 PM)Albert Chan Wrote: [ -> ] (01-21-2019 11:33 AM)Csaba Tizedes Wrote: [ -> ]I know what your topic is about, but show me somebody who need more than 10-15.

I remember one of my Professor fail any students 1 standard deviation below averages.

In other words, around 16% (1 in 6) will fail the class ...

I don't remember the class size, but it was definitely more than 15.

OK, but I am speaking about the "classical" engineering data collecting (you standing near a pump and set few measuring points with a valve and draw a diagram). And not a single data set, but measured data points, (X, Y) values.

For mapping a range no need more than 7-9, and 10-15 is far enough. And your measurement and analysis of data will not more precise if you make more measurement.

Because the site engineering is not a laboratory. But I am a mechanical engineer, maybe some electrical guy do it on another way. In my practice everything is dirty, sludge everywhere, compressors are few hundred kW's, blowers are higher than me, silos are few 1000 m3's, mass flows hundred tons per hours. In this case no need too much points.

...and for the single variable measurement, just calculate Student t-values divided by SQRT(number of measured data points). You will get something this:

Code:

`=====================================================================================`

Number of measured data (n): 2 3 4 5 6 7 8

Degrees of freedom (df): 1 2 3 4 5 6 7

=====================================================================================

Two sided probability (P): 0.95 | 8.98 2.48 1.59 1.24 1.05 | 0.92 0.84

| +-------------+

Two sided probability (P): 0.90 | 4.46 1.69 1.18 | 0.95 0.82 0.73 0.67

| +------+

Two sided probability (P): 0.85 | 2.95 1.32 | 0.96 0.80 0.69 0.62 0.57

=====================================================================================

And you can see clearly, if you want to reduce the variance of the measured data, you must to make at least n=7 measurement for two-sided probability P=0.95 and n=5 for P=0.90 (and n=4 for P=0.85).

Of course, this is not reflecting your exam-score example, because that is a measurement of 15+ independent variable and your professor generated from these measured values one property (the limit of acceptable score), and my example is about when you measures ONE variable 5-7-9... times. It is must to understand the difference. In your case it is required to collect AND store data if you want to draw the limit and later want to separate the measured values, but in my case this is not required. In my case the data collection purpose is increasing the reliability (decreasing variance).

(01-20-2019 02:30 PM)Maximilian Hohmann Wrote: [ -> ]If I would have to choose a single favorite calculator it will always be the Ti 59. Because of my personal attachment to this calculator: It helped me through my final year at school (where I was in the mathematical/natural science section) and through university (aerospace eng.).

Ah yes nostalgia. I used the Canon F-73P in high school and it's my favourite.

(01-20-2019 07:56 PM)pier4r Wrote: [ -> ]you can bump the "top 3 calculators" thread. http://www.hpmuseum.org/forum/thread-10433.html

Thanks pier4r, I will check it out.