Wrong date format when reading Excel file

Options
CGB
CGB Partner, L2 Designer, Dataiku DSS Core Designer, Dataiku DSS ML Practitioner, Dataiku DSS Core Concepts, Dataiku DSS Adv Designer, Registered Posts: 17 Partner

Hi,

I have some issue when I'm reading Excel files stored in folder.

I import 1 file by month in my folder and I create a dataset to read all files and obtain one stacked table as result. For some of them (but not all), the format is not read like it's stored :

  • In my Excel file, the column is store with dd/MM/yyyy format
  • When I read it with Dataiku, sometimes, it's MM/dd/yyyy format stored

So, at the end, I have mixed of format in my column and I can't use parsing on prepare recipe.

Any idea to correct this ? (I would like avoid to modify manually Excel files)

Thanks in advance

Best Answer

  • Turribeach
    Turribeach Dataiku DSS Core Designer, Neuron, Dataiku DSS Adv Designer, Registered, Neuron 2023 Posts: 1,757 Neuron
    Answer ✓
    Options

    The key in your post is "sometimes". This is a strong indication that your problem is on your Excel file. I am afraid that you really need to go back to your Excel file and fix your data. Most likely you have a date column which has a combination of dates stored as dates and dates stored as text. Use the techniques on the following link to fix your spreadsheet:

    https://www.macroption.com/excel-reverse-date/

    "I would like avoid to modify manually Excel files" => Then don't use Excel files, use CSV files which don't introduce a mixed format issue.

Answers

  • CGB
    CGB Partner, L2 Designer, Dataiku DSS Core Designer, Dataiku DSS ML Practitioner, Dataiku DSS Core Concepts, Dataiku DSS Adv Designer, Registered Posts: 17 Partner
    Options

    Hi @Turribeach
    ,

    Thanks for your answer !

    You're right, my problem is due to format storage in some Excel files : sometimes as date and sometimes as text.

    I use technique described on the link you suggest me and it works well.

    I made some tests on distinct storage, so I complete your answer if it could helps for other people :

    • if I choose to store as date with "DMY" year format, Dataiku always changes that on "MDY" format when I'm reading files and I can use option "Parse dates to ISO" to obtain directly a new column with Dataiku Date format.
    • if I choose to store as text, Dataiky keep my "DMY" format and it's read as string on Dataiku. To convert on Dataiku Date format, I need to use Prepare recipe ou code

    The first solution is the best one to avoid a second step executed in DSS Engine (parse Date isn't compatible with SQL)

    Regarding CSV file, I'm totally agreed with you : it's the best solution to avoid this issue. Unfortunately, these files are created by business team for their uses and not for mine ...

  • Turribeach
    Turribeach Dataiku DSS Core Designer, Neuron, Dataiku DSS Adv Designer, Registered, Neuron 2023 Posts: 1,757 Neuron
    Options

    Thanks for the update, I am sure this is a very common issue with Excel. To be honest I wouldn't blame the users, in this day and age for Microsoft to not fix this automatically when the user pastes/inputs the data is almost criminal. But they still charge everyone their 365 subscription no matter what...

Setup Info
    Tags
      Help me…