You will want to first import your data into Excel, whichever software you are using to analyze your data. If your data comes from www.keypress.com, you can export it directly into Fathom, but if you want to use Graphmatica you will still need to export it into Excel. We highly recommend that you also enter the data into your graphing calculator in order to use its regression analysis capabilities.
Modeling with Excel:
· Depending on the source of your data, you may be able to drag the URL address into Excel or copy and paste the data directly into Excel. Otherwise you may have to type in each data point directly in Excel.
· Your column headers should not have spaces in them if you intend to export your data into Fathom or Graphmatica.
· You will want your column headers going across the first row and each data entry in the rows below, one entry per row. If your data is in reverse form (with the headers going down the first column and each entry in one column), you can reverse them by doing the following:
1. Copy the data that you wish to transpose. Copy only the cells that you need, not the entire row/column. Include the column/row headers.
2. Choose the cell where you wish to display your transposed data. You may wish to move to a new worksheet in order to keep your data clear and unconfused.
3. Choose Edit Paste Special. In the dialogue box that opens, check the box for Transpose and click Okay.
· If your data is not in numerical order, you can rearrange it. Select your data, choose Data > Sort and then choose which column to sort by. Ascending will put the lowest values first; descending will start with the highest values.
· In Excel you can create a scatter plot of your data by choosing Insert > Chart. Select XY (Scatter) as the type of chart and click Next. You can change the title and also label your axes as you follow the steps of the Wizard.
· Once you have a scatter plot, you can add a regression equation. Select the chart, then choose Chart > Add Trendline. You can choose what type of regression line you want, then choose Options and to choose to display equation and R2 value.
Modeling with Fathom:
· First, you need to import your data from Excel into Fathom. Open Fathom and drag a new collection from the toolbar into the working area. Next, copy the data in Excel. Remember, your data needs to go down the columns, not across the rows, and your column headers cannot have spaces in them. Finally, choose Edit > Paste Cases. Drag a table from the toolbar into the working area so you can see your data and change any of the column headers if you so desire.
· Drag a graph from the toolbar into the working area. Determine which variable is the explanatory/independent variable and which is the response/dependent variable. Grab and drag the explanatory variable name from the table onto the x-axis of your graph, and drag the response variable name onto the y-axis. Fathom should automatically separate the data into a scatter plot. You can resize the graph like any other graphic, dragging the corners or edges.
· We recommend that you perform your regression analysis using your graphing calculator in order to be able to perform more regressions and to compare a linear regression to one done by Fathom. After you have found a regression equation, you can add its graph to your scatter plot. Right-click on your scatter plot and choose Plot Function. Enter your equation as a function of your independent/explanatory variable, not x – add the variable to your equation by clicking on Attributes button on the right side of the window. Click Ok to add the graph to your scatter plot.
· Fathom will calculate a linear regression line for you. Right-click on your graph and select Least-Squares Line. Fathom will also calculate a line based on medians. To show this line, right-click on your graph and select Median-Median Line. (To read about how this line is calculated, read about Median-Median Line in the help index.) Fathom will display the equation of either line at the bottom of the graph, along with the correlation coefficient.
· Fathom will only calculate a linear regression , so if your graph does not seem to show a linear trend and you want to compare other regression equations with those found on your calculator, you should consider using Graphmatica instead of Fathom. To investigate non-linear data, you could create an equation using sliders and try to manipulate it to match your data (ask Mrs. Powell or Ms. Sheehan for help), but your best bet would probably be to use Graphmatica.
Modeling with Graphmatica:
· You will first have to download the newest version of Graphmatica. Go to http://www8.pair.com/ksoft/ and click on “Download Graphmatica 2.0e” (in the middle of the page). Choose Open, double-click on “setup,” and then click Run. Choose to run as your ID number next to it and hit enter. Then choose Browse, select your H-drive (the one with your ID #), and type graphmatica into the box after H:/.
· To import your data into Graphmatica, copy the data in Excel. Remember, your data needs to go down the columns, not across the rows, and your column headers cannot have spaces in them. In Graphmatica, choose Edit > Paste Data Plot. You should see your data appear in the table on the left side of the window.
· Choose View > Grid Range to change the window size to fit your data. After doing so, you should see a point on your graph representing each of your data points.
· We recommend that you perform your regression analysis using your graphing calculator in order to be able to perform more regressions and to compare regression equations to ones done by Graphmatica. After you have found a regression equation, you can add its graph to your scatter plot. Type the equation into the white box between the toolbar and your graph. You will have to type the equation in the form “y =” some function of x. Then hit Enter or click on the toolbar icon that shows a pencil drawing a single line to display your graph.
· To calculate a regression equation, first click the Options button just above your data table. Choose the type of equation that you think might fit your data, including the degree if you select polynomial. Click Okay and then click the Curve Fit button to display the curve. Click on the curve and the equation will be displayed above the graph. Follow the same steps to experiment with other types of curves.
· Graphmatica does not calculate the correlation coefficient, so you will have to visually compare the curves to determine which best models the data. It will show a chi2 value, which is the sum of the squares of the differences between the curve and the actual y coordinates for all points - the smaller the chi2 value, the better the curve models the data. You can use this to help decide which curve best models your data, but do not completely rely on it; visual comparisons are equally important.