Confused on creating a normalized histogram (on excel)

[thepatient]

I have all the information I need, but I just need a bit of help on getting my data on a normalized histogram in excel.

First, if I'm not mistaking, a normalized histogram is just a normal histogram where it is roughly symmetrical about the curves centerline, is that correct?

I have a list of 25 data entries, and I have 6 bins, all on excel. I've been going through excel's help function and all over google to find out how to "normalize" a histogram but can't find anything helpful. XD Does anyone have an idea how I can do this?

I know that to make a normal histogram, you can go to "Data analysis", then choosing histogram, which allows you to choose input data values and bin's and automatically creates the histogram, but there is no option to normalize it.

Maybe I am completely off and don't really understand what normalizing means. :( Any help would be greatly appreciated. Thanks.

 

[Ray Vickson]

I have all the information I need, but I just need a bit of help on getting my data on a normalized histogram in excel.

First, if I'm not mistaking, a normalized histogram is just a normal histogram where it is roughly symmetrical about the curves centerline, is that correct?

I have a list of 25 data entries, and I have 6 bins, all on excel. I've been going through excel's help function and all over google to find out how to "normalize" a histogram but can't find anything helpful. XD Does anyone have an idea how I can do this?

I know that to make a normal histogram, you can go to "Data analysis", then choosing histogram, which allows you to choose input data values and bin's and automatically creates the histogram, but there is no option to normalize it.

Maybe I am completely off and don't really understand what normalizing means. :( Any help would be greatly appreciated. Thanks.

Google 'normalized histogram'. You will see that it means: re-scale the y-axis so the total area under the histogram = 1; that is, the frequencies sum to 1, so they are like probabilities. It has nothing to do with centerlines, or whatever. You can re-scale the y values yourself, then plot them as a bar chart.

RGV

 

[thepatient]

Hmm, but that sounds like the standard normal distribution, or the P(Z) function, which gives the probability. Are those two the same thing?

 

[Ray Vickson]

Hmm, but that sounds like the standard normal distribution, or the P(Z) function, which gives the probability. Are those two the same thing?

Absolutely NOT. A histogram describes measurements and may be representing a quantity that is very far from being normally distributed. For example, a histogram of data from an exponential distribution is about as far from the normal distribution as you can get. Still, we can normalize that data---but it still has nothing to do with the 'normal distribution'.

Do not be mislead by multiple usages of the term 'normal': we have normalized data (which may be non-normally distributed!), or we can have normally-distributed data (which might not be normalized!), etc.

RGV

 

[thepatient]

Hmmm... So basically we need to use the frequency and divide by the number of data points, and that will be our y-axis. Is that right? If we do that, then the integral under the histogram = 1.