Home / Brinton, Willard C. Graphic Methods for Presenting Facts. New York: The Engineering Magazine Company, 1914. Internet Archive identifier: cu31924032626792 (Cornell University Library copy). The first American textbook on what we now call data visualization. / Passage

Graphic Methods for Presenting Facts

Brinton, Willard C. Graphic Methods for Presenting Facts. New York: The Engineering Magazine Company, 1914. Internet Archive identifier: cu31924032626792 (Cornell University Library copy). The first American textbook on what we now call data visualization. 340 words

in Fig. 91 it would be following so far behind the periodic record curve that the ordinary reader would not realize that the moving-average curve is really up-to-date. For executive work, the object of a movingaverage curve is not so much to get a smooth curve as to show the average for the preceding year or other period of time considered. Under these circumstances it seems permissible to plot the moving-average curve as done in Fig. 215 instead of following the accepted method shown in Fig. 91.

In the last portion of Chapter VII some of the disadvantages of plotting curves on ordinary ruled co-ordinate paper were discussed. One of these disadvantages is due to the great difference in fluctuation with curves of small and of large numerical quantities placed near the bottom and near the top of a chart. In order to overcome this disadvantage, curves are often plotted on logarithmic paper. It seems desirable to point out that curves plotted on the curve cards described here are usually compared so that the disadvantages commonly found with curves plotted on ordinary co-ordinate paper are largely overcome. If only single curves are plotted on each curve card, and the zero line appears at the bottom of each card, then curves on different cards have approximately the same percentage scale. If two curve cards with different numerical scales are compared the comparison is much more accurate than would be the case if the two curves were plotted on a large sheet of paper to the same numerical vertical scale. The fact that the curves are all put into the same size of space on the curve cards causes them to have somewhere near the same percentage scale of height, even though the actual numerical scales may be widely different. Having all curves on the curve cards thus gives more accuracy and ease of reading than would be obtained if several curves were shown on large sheets of arithmetically ruled co-ordinate paper with one curve above the other to the same scale.