Need reference to algorithm that allows to decrease amount of points in chart dataset without losing meaning?

Need reference to algorithm (Java or C#) that allows to decrease amount of points in chart dataset without losing much of statistical/visual validity





For example dataset may have 5000 data points, but I need to reduce it to 500 (10%)

Need reference to algorithm that allows to decrease amount of points in chart dataset without losing meaning?
put in your java script 500 cd, type clrddds and press enter. you must delete all data sets charts so the comp has a clear idea of the visuality of the items. only the statistics from 3 months ago will show. i dont know why b ut the visual must be set to 15% for you to see the c drives options,
Reply:You need to filter the data. For a 10:1 reduction in the data record, you need to apply a digital anti-aliasing filter. Then decimation in time to cull the unwanted interstitial points.





For example, in your example, take your 5000 data points and run it through a low pass filter with a digital bandwidth of 0.09*pi radians and with a stop band attenuation of at least 80dB, then take every 10th point and put it in your new array. You will end up with 500 points that preserves as much information about the original record as can be achieved.





Another way to do this is to take an FFT of the points and apply and inverse-FFT to the fraction that constitutes the lowest frequencies. Easier to do when working with record lengths on powers of 2





For example, take 8192 points, perform an FFT on them. Keep the first 512 points and the last 512 points and mash them together to make a 1024 point record. Then shove that through an inverse-FFT to get a 1024 point compressed record of the original 8192 point record.
Reply:not into java or c# but i can di it in php
Reply:In the real world, we would sample the randomly sample the data set. In most cases, the sampled data set would give "nearly" the same results as the full data set.





Unless you have duplicate data points, or a simplistic linear data relationship among your original data points, you will lose some meaning by reducing the sample size -- no matter how you choose the data points.