Day of week in custom dimension?

Hi, is it possible to extract day of week or day of week index in custom dimensions or table calculations? I don’t see anything in https://docs.looker.com/exploring-data/creating-looker-expressions/looker-functions-and-operators?version=5.22. Thanks!

2 Likes

Hi @ravimody ,

We can get the day of week index in a table calculation by using a combination of the diff_days and mod functions:

mod(diff_days(date(2008,01,01), ${orders.created_date}) + 1, 7)

This will return the day of week index with Monday being 0.

If using Redshift or Postgres, this formula will need to be modified to include the round function:

mod(round(diff_days(date(2008,01,01), ${orders.created_date}) + 1), 7)


We can also get the day of week using this same formula and writing in the days for each number like this:

index(
  list("Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday", "Sunday"), 
  mod(diff_days(date(2008,01,01), ${orders.created_date}) + 1, 7)
)

Hope this helps!

Ryan

1 Like

This is great Ryan, thanks!

You can also specify a timeframe in your table View to add “day of week” as a field or filter provided that your database is referencing a timestamp.

Example:

  dimension_group: created {
    label: "Page Creation"
    type: time
    timeframes: [
      raw,
      hour_of_day,
      time,
      date,
      day_of_week,
      week_of_year,
      month,
      month_name,
      quarter,
      year,
      day_of_month
    ]
    sql: ${TABLE}.created_at ;;
  }

1 Like

ryan.dunlavy:

We can also get the day of week using this same formula and writing in the days for each number like this:> > auto> index(> list("Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday", "Sunday"), > mod(diff_days(date(2008,01,01), ${orders.created_date}) + 1, 7)> )> >

This works great for me except I’m getting null values instead of Sundays. I expect that’s because the first day of the week is set to Sunday in our instance of Looker. Is there a way to adapt this formula to allow for that?

coalesce(
   index(
      list("Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday", "Sunday"), 
     mod(diff_days(date(2008,01,01), ${orders.created_date}) + 1, 7)
   ) , "Sunday")
2 Likes

I found a different way to do this. I wanted to highlight Weekends in our visualizations. I added the day_of_week date part to the date LookML, and to the table. Then created a custom calculation that was just:
if ( ${order.created_day_of_week} = "Saturday" OR ${order.created_day_of_week} = "Sunday", "Weekend", "" )

1 Like

If I try to do pivot where this field = “Monday”, it returns null.

pivot_where(${calculation_1}=“Monday”,${v_sec_cdrs.interaction_count})

Calculation 1 being the change from date to weekday

index(

  list("Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday", "Sunday"), 

  mod(diff_days(date(1900,01,01), ${dw_activity.completed_on_date}), 7)+1

)

I used this instead, the +1 should be outside of the mod so it forces to start at 1 and not 0. However, i found that starting at 2008 did not make the date accurate vs starting at 1900, unsure why but it would be one day off.

@ryan.dunlavy the modulus calculation you provided is 1 day off for me as well, as @genghisk also mentioned. Perhaps you can update your comment, so less people copy&paste your solution.

1 Like