Regression 2.0.1 Read Me Regression 2.0 was designed for physics related lab computations. Many experiments in laboratory conditions require comparisons between two sets of variables. Often long logarithmic computations are necessary to derive an equation. Regression 2.0 was designed to eliminate this step by automatically providing power, exponential, logarithmic, and linear relationships. Requirements - Any Mac with System 7.X * If you find any other problems, Iąd like to know. Regression was tested in a high-school classroom for about three months. They didnąt report any problems, but Iąm not counting on that. Regression will run in black and white, but the splash screen and about box will look terrible! Features: € Variable sets can be saved € Supports Drag and Drop € Five Regression dialog boxes can be open at once € Computes relations of: - Power - Logarithmic - Exponential - Linear € Can check variables: i.e. enter X, get Y or vice-versa € Improved variable editing from 1.0 € Easy to use compared to large commercial packages € Computer degree of error for a set What is Regression? Regression 2.0 was originally programmed by request for a physics class. The instructor wanted a small, simple to use program that wouldnąt be difficult to teach to students. There are many large, commercial programs out there, but many of them have features that are, for the most part, unneeded. Often these programs take quite a while to master. While Regression was designed for a limited range of functions, none-the-less it filled the requirements for the physics class. It was not designed to eliminate all the work for students, just the tedious mathematical part. Also, the commercial packages can be quite expensive, especially if you donąt want all of the features. When we started Regression 2.0, we wanted to maintain simplicity. Most labs donąt have big computers, so Regression can run with about 400k. It can also easily fit on a disk. I havenąt had the patience to try a really long data set, but if you enter sets larger than 100 or more, increase the memory requirements by about 50k. To do this, single-click on the Regression 2.0 program icon and press Command-I. In the Memory Requirements section, change the 400 in the Preferred Size box to 450. This is probably overkill, but rather safe than sorry. Note on the mathematical engine: The physics instructor supplied the internal mathematical computer code, but these equations were derived from an older programming language on an ancient TRS-80. Most of the variables were labeled A, B, etc., and this was terribly uninformative. Because of the complexity used in programming the mathematical equations, we simply translated them. Due to this, we are not fully aware of the limitations of the engine. However, this is what Iąve discovered so far: 1.) The engine shows no problems for all positive real numbers. Real numbers are valid to 1.7 x 10 38, and strings are limited to 255 characters. In a college/high school laboratory environment, no one should get even close to these values. 2.) It is not restricted to integers. Decimals are perfectly valid. 3.) It does not handle negative numbers. The instructor didnąt see this as a big limitation. After all, if you start at zero, you shouldnąt have to measure anything negatively. I allowed Regression to have accept negatives because under certain circumstances negatives produce results. In my experience in physics classes, I saw no need to have negative numbers. Note: this only includes the X and Y set. All of the other functions involve standard Pascal routines. Check should not be affected. Regression 2.0 Instructions Most commands are standard, so unless you are unfamiliar with the MacOS you should have no problem. To enter a new set from scratch, select New from the File menu. Once the main Regression window appears, click on Input Set or press Enter. From here you can enter both the X and Y variables. Enter only valid numbers between 1 and 255 digits. To proceed to the next set press enter or click on Next or >>. When finished simply click on Done. To scroll through the list, << button takes you back and >> takes you forward. The numeric pad was designed to facilitate the entering process. Clear doubles as the tab key normally would. Tab works as well, but its all the way at the other end of the keyboard. Once youąve entered a set, it will appear in the list box. If you see any erroneous variable, click on it in the list box and then click on edit. Or, just double-click on the variable and it will take you straight to the enter variable dialog box. You can also insert sets before and after the row that is currently selected by using their respective buttons. Check was included so you could enter an arbitrary X or Y and get the other variable. This is useful to test accuracy or filling in missing data. You can switch between equation types of equations by selecting the corresponding radio buttons at the top of the window. The equations for each of these is given just above the results box. Save and open work like any other program. However, you can double-click on a Regression data set icon and itąll load the program and the set. Drag and drop works, too. You can open up to five sets at a time by dragging the icons onto the Regression program icon. The box in the lower left-hand corner displays the results of the calculation. Accuracy is very useful in determining how close each set is to the equation. The range goes from 0 to 1, with 1 being perfectly correct. A score of .99999346 is very good for a college or high school lab. .8900885 is relatively poor. The better the value, the more you can trust the values given for M and b. M is the slope and b is the Y-Intercept. The relative relationship of M and b are given in the equation. In other words, if your data was accurate (usually better than .9999 +), then the check will be accurate as well as the equation the program provides. This is one of the best ways to verify the validity of the data. For example, if you know 100 grams, X, pulls 250 grams, Y, on a certain pulley, but the equationąs accuracy is < .9999 and it gives you 325 for Y you know that your data was not accurate. Regression 2.0 Registration Regression 2.0 is Shareware. If you use the program for more than 30 days, there is a $15 registration fee. Please send it to: Aaron S. Knapp 6344 St. Rt. 380 Wilmington, OH 45177 (Checks payable to Aaron S. Knapp) We welcome any comments or suggestions you may have concerning Regression 2.0. If you find any bugs, report them and we will post a revised version as soon as possible. © 1996 BattleAxe Software. All Rights Reserved.