This week I thought it would be interesting to explore some real data. I have downloaded this from the Irish government data website.
It tracks Sea Surface temperature since 1958 at a place called Malin Head (northerly point of Ireland).
Here is where I clean it up and I thought it would be interesting to see what AI gives me around insights etc.
So I began by asking Co-pilot to clean up the data i.e.
Split the column ( ) into Date and time.
Remove redundant columns e.g. I don’t need blank columns or the station_Id (it’s Malin Head for all 350,000 plus rows)
Remove the metadata at the top and bottom.
I decided that what I wanted to show was a summary of the data with a line chart and trend line that I could analyse by Instrument name and to be able to drill into specific years.
What I noticed
I ended up just removing the metadata myself – actually quicker than Co-Pilot
Then I had to:
- Save the file to OneDrive
- Make sure it was saved as an Excel workbook
- Converted to a table
- Autosave turned on
I then got it to do the clean-up:
Remove unwanted and blank columns and split the Date column into Date/Time. However I forgot to tell it to format the date as dd/mm/yyyy and I had to get it to re-do it. I have shown some of the techniques I used to check and format as date in the video.
Thoughts
Co-pilot is really helpful around cleaning up the data. However, I wasn’t so happy with what it gave me but on the other hand maybe I need to work on my prompting skills – so need to think about that.
Off screen I experimented with getting Co-Pilot to create what I wanted but it gave it to me in a PDF – not what I wanted. However it did give me quite clear instructions on how to do what I wanted.
This is the link to the raw data file
Here is the starting file here – ready to use with Co-Pilot
Here is the finished file here
Here is the video..