# How to calculate percentiles 00:00:00,000 --> 00:00:08,101 Excel has a function called PERCENTILE DOTTING, which is extremely useful when we want to identify deciles in the data.   00:00:08,101 --> 00:00:23,384 We are assuming temperatures in the first to third decile are cold and temperatures in the seventh decile or above are hot for the years 1951 to 1980. We want to find out what those temperatures anomalies would be. 00:00:23,384 --> 00:00:32,267 First type a label into column U to the right of your data so that when you calculate the numbers other people will know what they are. 00:00:32,267 --> 00:00:38,785 If you've just done the previous data exercise you'll have filtered your data so it shows only 1951 to 1980.  00:00:38,785 --> 00:00:42,396 Remember the rest of the data is there, it's just hidden. 00:00:42,396 --> 00:00:50,939 If you've not filtered the data, go back to the previous exercise and filter it again. The previous walkthrough tells you how to do it. 00:00:50,939 --> 00:00:55,860 Click on the cell to the right of the label for the third decile and type the function name. 00:00:55,860 --> 00:01:13,294 That's =PERCENTILE.INC and an open bracket now select all the cells containing the temperature anomaly values that you want that's columns B to M. 00:01:13,294 --> 00:01:23,391 Then type a comma and then the decile number. We want the third decile and so in Excel we need to type 0.3. 00:01:23,391 --> 00:01:33,360 Close the bracket, hit Enter and the third decimal value appears in the cell. Do the same for the cell below it. 00:01:33,360 --> 00:01:49,832 Type the function name, open a bracket, select your data, comma and 0.7 for the 70th percentile.   00:01:49,832 --> 00:02:04,174 For the data that we've used the number is 0.11 for the 70th percentile. Now what does mean? It means that for this data, 30% of the observations are greater than this number. 00:02:04,174 --> 00:02:16,233 As a way to check you haven't made a typing error, just take a moment to look at the data you used, just to check that your number seems realistic. You avoid a lot of errors this way.