I want to predict values.i have established a 95% C.Interval,but the range of my data exceeds my C.I,need hel?

Question

i have generated a lot of measured data over time.now i want to observe trends in my data values and correctly predict some values without measuring them.however, the range (max-min) of my data exceeds my C.I. , the same thing applies to my mean+ or- 1s.d. should i just forget it or can my C.I. help me achieve this

Answer 1

It would be helpful if you gave a little more detail about what you're trying to do specifically.

If you're trying to predict the stock market, you should just forget it.  It won't work.

The confidence interval is the probability of a given measurement occuring assuming that your data is normally distributed.  If your data drastically departs from normal.  Also, if you have a lot of data, then of course there is some probability that due to noise it will not lie within  your C.I.  .. that's why range (max-min) is not a particularly useful statistical concept.

Also ... if things seem really off, recheck the definition of confidence interval that your program is using.  Sometimes people use confidence interval and significance level interchangeably -- make sure when you are using 0.95 you don't really mean to use 0.05.

Answer 2

Not sure what a CI is. But in anycase, open up an excel spreadsheet. Now paste all that measured data in a column. If you wanna see the trend in which those samples fell, use the frequency function, that will count all the samples falling into the range intervals you set in the second column. Just check frequency in the help. If some samples go out of your range, it will still handle it nicely. It is very easy to plot the output of this function as well, as it will make a new column with all the values.

Now for mean, just use mean function on your data column. Same goes for your median. Excel is your friend (just a note, if you have Open Office, look out, some versions have a bug in the frequency function which wrongly calculates the values.)