Exploring & curating data
Dashboards, Looks, Visualizations— it all happens here
I like to track Quarter vs Last type trends. EG in the following chart: The challenge I have in accomplishing this in Looker is that I have a date field, but I don’t have a “dayOfQuarterIndex” or “weekOfQuarterIndex” field. The data table I would need to create would look like this: Does anyone know how to create the quarter’s week index from a “Created Date” field? (The created date field is in standard iso format YYYY-MM-DD with a timezone offset). Thanks, Wyatt
Dashboard has 3 looks/ tiles. I’m applying the same filter (Company Name) to update all 3 tiles, so I can see the numbers for all tiles that apply to that specific company in the filter. When I export this dashboard (either via Downloading as PDF of scheduling an email with PDF attached or Inline image) I do NOT want the filter to appear in the visualization of the dashboard. The only way I see possible is going to settings, choose “Edit Embed Settings”, then turning OFF the switcher “Show filters”. When I click on “Preview Embed”, indeed the Filter is not showing. However, when I send the dashboard as scheduled email - either as PDF attached or inline - the filter is showing. On the PDF version, it goes on top of dashboard. On inline version, it shows on each tile. Any ideas on how to make the filter truly invisible when Dashboard is exported via email?
This is a technique for creating single value results with multiple measures. Instead of making a single query using filters and a single measure or dimension and a single value visualization for each result, return an entire row with every single value desired: Now from this single query, we can create a single value element that contains multiple measures of data by hiding the dimensions and measures and concatenating them, be sure to hide the fields after concatenating. This will reduce the load time used in running parallel queries on the database which uses the same WHERE filters. It can also help when designing dashboards to reduce overall tiles by consolidating similar metrics.
Hi - Can I make a request to enhance the conditional formatting options for the table viz? Specifically I need: ***Ability to map a rule to a column so I can have multiple rules per table and column specific color coding This is very import to how I manage my business and has huge business value. PS - I am a business user and can’t code or modify HTML so that option is not on the table.
Is it possible to setup a look for users to use without it running by default? For example, I’d like to have a predefined query that allows them to “fill in the blanks” and click Run. I see that “explore from a look” almost gets me there, but it wants to run automatically.
I have a use case where I have duplicate records but difference dates and I want to get a count of Unique Records by only selecting the first recorded Record For example, Data Set: UID-Date-Value A 6/12 Registered A 6/12 Registered A 6/13 Registered B 6/14 Registered Want to Return: UID-Date-Value A 6/12 Registered B 6/14 Registered Count = 2 How can I do this in Looker?
If you haven’t already read my post on using the beta distribution for an A/B test, that might be a good warm-up read. (The disclaimers from that post still apply 😉 ) In that example, the data took the form of a proportion - a number of successes out of a number of trials. What do you do if you want to instead compare rates of events over intervals? For example, user signups, error events, or web traffic over time. Well, in the A/B testing example, the data points can be thought of as being taken from a Bernoulli distribution: either yes or no with a given probability. That probability is in turn modeled by the conjugate prior of the Bernoulli distributions, the beta distributions, so we used the beta_inv table calc. In the present scenario, the data points can be thought of as being taken from a Poisson distribution: its possible values are whole positive numbers representing a number of events happening in an interval assuming those events occur independently at a given rate.
Hi - relatively new with Looker but old SQL person. This may be a simple question. I have users that are trying to build a Look that lists out rows from a Master view that lack any activity in a transaction view. To be specific, they have a master view of students (one record per student), and a transaction view of job applications. I have measures in the model to count applications, but because of the way the JOIN works, it is only showing those students with at least one application transaction. In SQL I would simply “select … from students where studentID NOT in (select studentID from applications)” How can a user accomplish this in a Look? Thanks!
To start, we’re currently on Looker release 4.14, and I am an admin user with “download_without_limits” With a data set that has a zipcode type (I do not have reliable Lat/Long to use the location type), I am building an US-map visualization with dimensions of zipcode and one other, and the measure being a simple count of rows. The map works great right now, but the data set is limited to 5,000 points, and I have a full set of about 7,500 (from my SQL running directly against the database). Does anyone know a way to move past this limit?
Hi, when I tried to import the LookML dashboard into a new Space, in this new version of the looker, the Looks associated to the dashboard were not imported. In the previous version, the Looks were imported together with the LookML Dashboards. Is there any change in the API or this is a bug?
Hello, Lookers! Is auto-scrolling and/or auto-changing dashboards in your Product Team roadmap? (like Klipfolio’s tv mode, where selected dashboards are loaded every x minutes) If not, anyone recommends a working plugin that can, at least, auto-scroll up and down the dashboards? We are trying to implement some monitoring screens and this feature would be really helpful!
Hi, I show strange behavior for Calculated fields in Look (calculated fields get dropped from the Look/explore from here if add additional measures on the Look. I had few measures and one calculated field on the Look. I added 2 more measures and moved them at appropriate positions but it dropped calculated field which was there when I opened look using explore from here option. I have to recreate calculated field again for all the look with similar modification. And this getting too much time confusing as have to create all calculations again. Looker version: 4.22
Hi all, I’d like to build XmR charts using Looker, but have not yet succeeded in doing this. I’ve some time-series data from the first of January up to today, with new data being added on a daily basis. The dataset includes one value for my performance measure, per day. For the XmR chart I need to calculate the mean and standard deviation for a subset of the dataset, the control data (14 fixed, consecutive values of my performance measure in the dataset). However, I’m not able to engineer table calculations such that it creates a measure which, for every date, computes the mean across the same time window. For example (see mock-up data), I’d like to calculate the mean and standard deviation for the period 2017-08-06 until 2017-08-19 and subsequently include this as a measure, where this value should remain constant (and the time window should stay fixed), also if new dates are added to the result set. Anyone that has ideas how to solve this problem? Thanks!
I have few measures on the Look. Now i want to apply conditional formating to measures based on certain rules. Now I only want to apply certain rules to few measures and others on other measures. But Looker only allows to whether to apply or not apply rules, is there a way to have customize formating for each measures at the visulization level? Also is it possible to replace measure values with String if data is Null using conditional formating or calculatee field. I have percentage measure on the look and want to change it to ‘Not Tracked’ if value of the measure is null else display percentage value. Looker is not allowing to have boolean, string, int conversion using if() as this is not the valida case for the SQL (I believe). Is there a way to achieve this type of formating at the Look Visulization level?
I want to create comparison chart like it should pull most recent month of year and compare it same month of previous year. Eg My recent month is Aug 2017 it should pull the same and compare it with Aug 2016 (it should display bar graph of Aug 17 and Aug 16)
Disclaimer: The statistical methods in this post are Bayesian. I strongly prefer Bayesian statistics over frequentist statistics, and I think you should too. But, that’s a post in and of itself, and there are plenty of resources easily found on the internet that do it better than I could, so I will leave it at this - if you do any statistics, but aren’t yet familiar with Bayesian statistics, please take some time to read up on them! Looker 4.22 ships with a ton of statistical functions, including several inverse distribution functions for conjugate priors. If that didn’t make any sense and you just want to run an A/B test, worry not & read on. Let’s say you have randomly assigned visitors to one of several variants, and collected some data like this: Variant | Visitors | Converted ------------------------------------------ Control | 519 | 34 Test-A | 217 | 23 Test-B | 224 | 19 What can we conclude from this using Bayesian statistics? We can build credi
We’re struggling to figure out a solution to identify when one or more events are low or missing from a table. Let’s say we want to confirm that these events all have 1,000+ records per day: event_a event_b event_c … event_z We filter eventname for events a through z with an event_count of less than 1000, but our alert will only trigger if there’s 1 to 999 events for a given eventname. If event_a has 0 events, it won’t trigger as an alert. We can’t think of a dynamic solution short of setting up one alert per eventname, which is costly (bigquery) and inefficient, or one measure per eventname, which is just inefficient. We could create a “count of eventnames” and always confirm the result is the number of events we’re filtering for, but that won’t consider the number of events per eventname. Any other ideas?
Already have an account? Login
Login to the community
No account yet? Create an account
Enter your username or e-mail address. We'll send you an e-mail with instructions to reset your password.