<p>hey, so we're on this new topic in class and i haven't really been paying attention.</p>
<p>we use workshop statistics, which pretty much gives problems w/o much explanation.</p>
<p>if someone can explain to me which cases require a transformation of explanatory + response variables (log x, y) (x, log y) (log x, log y) etc, it would be greatly appreciated</p>
<p>I think your talking about exponential and power regression. If so, then u need to transform the actual "curved" data into a linear line. For exponential regression, u have to graph the original data (x,y) . Then, u find the log of only the y-variables. then, u graph (x, log y). Use linear regression to find the slope and y-intercept for the line of best fit. Graph the line. in the Y= screen, set one of the Ys to 10^(Exponential Regression). This should give u a line that will fit the original data (x,y). Now u can predict future y-values and inter-domain y-values.</p>
<p>For power regression, you have to graph the original data (x,y) . However, this time u have to find the graph the log of both x and y (log x, log y). Then you do another line regression. plug in the equation. Then do an power regression and then graph that line against the original data (x,y). </p>
<p>I am explaining this presuming that you have a TI-83 or higher model for a calculator. </p>
<p>Also, explanatory is the independent variable (x) and response is the dependent variable (y).</p>
<p>Hope this helps. ;)</p>