Gold Study

Question 1

When I add many fields to my rows shelf, some rows start going vertical. As I add more, the labels start getting concatenated. Can I fix this?

Answer 1

You can adjust Tableau from taking these actions by setting default table options. (Analysis  Table Layout  Advanced, and increase the number of row and horizontal row labels.)ALSO: You can perform some of the following actions:• Resize the label rows (click-drag) to provide more room• Choose to “Rotate” the labels (right-click a label)• Edit the aliases of the label values to make them shorter.• Format the font for the labels to make them smaller• Un-Choose “Show Header” for some field labels• Choose “Hide Field Labels for Rows”• Don’t place so many rows onto your sheet – it’s not best-practice for visual analysis!!

Question 2

How does unchecking ‘AnalysisAggregate Measures’ differ from converting a measure to a dimension?

Answer 2

A dimension returns all distinct values of the field. “Dis-aggregate” means that you want to return each row from the underlying database. Example: 1,1,1,1,2,3. When set to a dimension there would be 3 marks, as a disaggregate measure, 6 marks.This becomes very important when using reference lines, trend lines or other secondary calcs. An ‘average’ ref line on the dimension would yield a value of 2, but on the disaggregate data the value would be 1.5.

Question 3

What is an ‘Attribute’ (ATTR)? How does it differ from dimensions and measures?

Answer 3

An attribute is an aggregation that can be applied to Dimensions. This enables the dimensions to be removed from the level of detail, but still be displayed. The functionality is useful for excluding a dimension from a table calc or allowing dimensions from secondary data sources. Returns the value of the dimension if it only has a single value for all rows in the group, otherwise it displays an asterisk () character. Null values are ignored.An attribute is equivalent to the formula: IF MAX([Field])=MIN([Field]) THEN MAX([Field]) ELSE“” END

Question 4

Can I connect to a stored procedure?

Answer 4

Sometimes, but this is not often needed. Due to Tableau’s advanced calculation, filtering and security capabilities, most customers find that they can accomplish more with tableau connecting to raw data than by executing a stored procedure.The stored procedure or User defined function (UDF) must return a table. The reason for this is that all subsequent actions in Tableau will be sent to the database as queries in a ‘sub select’. For Example:Select [Region] From ([my stored procedure result]). Most databases don’t support this. When they do, this effectively means that the stored procedure is executed every time an action is taken in Tableau. This typically defeats the benefits of a stored procedure. A much better approach is to first execute the stored procedure into a table or file and then connect Tableau to the results.

Question 5

Does Tableau have any API’s or scripting?

Answer 5

“What is it that you are trying to accomplish?” We have embedding and URL actions that solve a lot of these needs. Other items are often solved with existing functionality such as table calcs or actions that will help satisfy their requirements. We also have a Javascript API.

Question 6

Your website mentions the VizQL language. How do I program in this or modify the VizQL?

Answer 6

Currently the VizQL is generated only by using the Tableau Desktop interface. It can be manipulated through the Tableau Server web interface by Interactors. It is not possible to see the VizQL or create it directly.

Question 7

How do I group thousands of items together?

Answer 7

Best chance is to use a calculated field with a programmatic condition in these “thousands of items” such that the cardinality is low (less than 100). For measures, rounding using a calc field is useful. For dimensions, rounding using or LEFT or RIGHT functions, etc…

Question 8

Can I save a filter for reuse? For example I manage 12 of the 5000 products we have.

Answer 8

Yes. Create the filter, and then choose “Create Set” from the filter’s context menu

Question 9

If I have a calculated field that includes members from two blended data sources, will the results be included when I create an extract?

Answer 9

Since extracts are specific to each data source, the results from the calculated field will not be included in the data extract (i.e. it will not be optimized). However, the calculation will still work fine.

Question 10

How do I move a sheet from one Tableau workbook to another?

Answer 10

Bookmarks. Save one from the source workbook. Open the destination book, choose that bookmark, save the file. (Note: you cannot bookmark a dashboard)

Question 11

Can I move dashboards from one workbook to another?

Answer 11

No.

Question 12

What is the polygon mark type and how do I use it?

Answer 12

Polygon mark type is advanced and used to describe geographic areas and other areas of measurement. They typically require coordinate data in the underlying database. Armed with the right set of coordinate data, polygon marks – in conjunction with the PATH shelf – can describe two-dimensional areas. Any time you have all of the coordinate data to describe ANY area (state, election regions, etc), you can accurately use the polygon mark type to show these boundaries.

Question 13

What are the two main types of Joins?

Answer 13

Joins require a condition to evaluate how a row in one table matches a row or set of rows in another table. The fields used in this condition are called ‘key’ fields.Inner join: a join of two or more tables that keeps only that rows where the key is common to both tablesOuter join: is further defined as left outer join or right outer join. These includes all rows from the table on the left (or right) side of the join condition regardless of whether a match was found on the right (or left) side.

Question 14

I created a join and most of my data is missing. How do I fix this?

Answer 14

An outer join will always include the records of the primary table (i.e. left outer join will always include data from the ‘left’ table, even if no match is found in the right table.) An example is a transaction header table which is related to a transaction line-item table. For dimensional analysis purposes, you would want all header rows to be returned, even if there was not always a specific line-item related to it.Select * from TransactionHeader LEFT OUTER JOIN TransactionLineItemsOn TransactionHeader.LineItemID = TransactionLineItems.LineItemID

Question 15

What is Join Culling?

Answer 15

Because joins cost time and resources to process on the database server, we don’t really want to use every join that we declared in our data source at all times. Join Culling allows you to skip the join when the query does not require it and provides increased performance. To do this, you must have relationships setup in your database between the tables you’ll be joining in Tableau. Inner Joins will work best for Join Culling.

Question 16

How do I do Sparklines?

Answer 16

Create a line chart.Edit the axis to not include zero.Hide the axisMake the rows really tiny

Question 17

How do I create a chart the shows the budget compared to actual?

Answer 17

Try a bullet graph first. IF that doesn’t satisfy, then:Color Answer: Steps: add “budgeted” to ROW, add a date measure to column. Then, add second measure (e.g. “actual”) to Y-axis (i.e. the row axis where “budgeted” was already placed). Tableau intelligently determined that COLOR is the best-practice in this case.Side by Side Answer: Follow the steps for Color Answer, and drag Measure Names to the Column Shelf next to your Date. This allows different measures to appear side by side.

Question 18

How do I show the rank of my products?

Answer 18

1. Create a calculated field: index()2. Add this to your sheet and change it to discrete.

Question 19

Can I create a chart that shows the trailing 12 months revenue?

Answer 19

Yes, this is easy to do in Tableau. You can use a table calculation for this, as well, you can use the built-in totals feature. If you have “month” as a dimension you can turn on grand totals against this dimension.Answer one: use relative date filters to filter a single sales bar down to the last 12 months (don’t add date to the columns). Answer two: Use a Table calc to use a Moving Calculation for the sum of the last 12 Months. This one is even better as this is simple for Tableau, and hard for other products.

Question 20

Can I create Control Charts?

Answer 20

Yes. Control charts are often used in SPC (Statistical Process Control) and six sigma (6σ) efforts or other quality control. It is basically a time series analysis with trends and reference lines. Tableau makes them very robust with one exception: We can exclude the outliers completely, but cannot exclude them from the reference line calculations while still displaying them on the chart.See the knowledge base for more details.

Question 21

Can I do a Pareto Analysis?

Answer 21

Yes (this is the quick way, a more complete answer is in the knowledge base): Place a dimension on the column shelf. Sort it by descending.Place a measure on the rows shelf. Duplicate it so there are two measures.Set the 2nd one to be a Running Total with a secondary calculation of Percent of Total.Dual Axis these two measures.Set the first measure to be a Bar and the second measure to be a Line.

Question 22

I have a view with MY(Order Date) on Columns, SUM(Sales) on Rows, with three different colored lines – one for each Product Category. I want to animate the lines over time, and show the history so I can get a view like the following where the lines will appear for each month:

Answer 22

Drag Date to the Pages shelf and to columns. In order to get the lines to produce over time, the Mark type needs to be changed from Line to some other option, such as Shape, Circle, even text. Then check Show History on the Pages card, access the drop-down menu, Show History for All and select Trails.

Question 23

Can I use dimensions from the secondary datasource to roll-up my primary data? How can I create a primary group from a secondary data source?

Answer 23

Yes. Kind of.An example is if I have sales and people in my primary and people and their territory assignments in the secondary. How do I roll-up my sales (primary) by territory (secondary)?Once your data blended view is complete you can right click on your field you wish to group from your secondary source and select “Create Primary Group.” This will populate your primary data source with a grouped field based on values shared in both sources.

Question 24

Setup a view that displays only a list customers whom purchased from BOTH Furniture and Office Supplies in the past.

Answer 24

Since no row of record contains information on about a customer whom purchased on multiple product categories, this is a filter based on aggregated results – this suggests using the “conditions” tab in filter on [Customer Name]Use the formula: (MAX(IIF([Product Category]=”Furniture”,1,0))+MAX(IIF([Product Category]=”Office Supplies”,1,0)))=2How this works: • MAX(IIF([Product Category]=”Furniture”,1,0))Show me the value “1” if Customer ever had bought something from Furniture• MAX(IIF([Product Category]=”Office Supplies”,1,0))Show me the value “1” if Customer ever had bought something from Office Supplies• (MAX(IIF([Product Category]=”Furniture”,1,0))+MAX(IIF([Product Category]=”Office Supplies”,1,0)))=2Only show me if both of the above formulas returned “1” or another words 1+1=2

Question 25

Can I use Tableau to discover which products were frequently purchased together?

Answer 25

Yes. This is commonly referred to as a Market Basket Analysis. First, identify the unique identifier (Customer name, ID number, Business, etc.) and the category in question(Product category, Business segment, Academic Course, etc.). Self join tables approach: Go to Edit Tables on the data connection and choose to Add a Table. Select the table with the data we will be needing and choose to Join the Table. Add two join clauses: Join the identifier to itself, and join the category in question to itself where the relationship is (not equal).

Question 26

How about the top 5 products that I sold this year?

Answer 26

Yes. Filter on current year, then add this filter to the “context”. Reason: context is created prior to filtering to the top 5.

Question 27

How about the top 5 in each region?

Answer 27

Yes. Create an index() calculation and add it to the sheet. Edit it to restart for each region.

Question 28

What about how the top 10 products and regions are performing?

Answer 28

Yes. Make a set of product and region, place this new set on row, followed by product and region. Sort on the set. Then hide headers:

Question 29

How do I make a box plot?

Answer 29

Many customers prefer our box plots to the standard definition because we will show you all outliers and plot density throughout the range, not just the whiskers (though you can get that as well).See the knowledge base for examples.

Question 30

Can I find the purchase history of customers that purchased this month?

Answer 30

1. Manually filtering2. Dynamic sets from dimensions3. Manual sets from marks/members4. See also: cohort analysis.

Question 31

Every day, I want to open Tableau and see how yesterday’s top 3 sales people have performed over the last 6 months.

Answer 31

Similar to above, use dynamic sets. Once created, this set can be a filter and be used against any date range. You need to peg it using “add to context”.

Question 32

How do I apply weights when working with my survey data?

Answer 32

If you have a weight field in your data set you can apply this weight with a simple calculated field. [Weight]*[Measure].

Question 33

How do I find the average sale per order?

Answer 33

Calculated field. Given a dimension “Order” and a measure “line item sale amount”Sum([line item sale amount])/Countd(Order)

Question 34

Tableau only created equal sized bins, but I want mine to be 0-10, 10-50, 50-100, 100-1000, and greater than 1000.

Answer 34

Instead of using built in binning, use a calculated field with an IF statement. Something like:IF [Sales] > 10000 THEN “LARGE” ELSEIF [Sales] > 5000 THEN “MEDIUM” ELSE “SMALL” END

Question 35

Your bins are great! But I have just a couple of things that are bigger than 1000, how do just create a bin that is >1000 and not have to define all the intermediary bins?

Answer 35

1. IF [Sales] > 5000 THEN 5000 ELSE [Sales] END2. Then, bin this calculated field3. Place on shelf4. Edit alias for that last group (e.g. ‘>= 5000’)

Question 36

Is there any way that we can dynamically change the size of our bins?

Answer 36

Yes, Tableau makes this very easy to do by allowing you to create parameters directly in the bin menu.

Question 37

I have a date field that looks like 20070326. Tableau doesn’t recognize it as a number. How do I make it into a date field? Bonus, I have 200703?

Answer 37

Calculated field. Examples assume the source is of type “string” (otherwise, wrap str() around these):DATE(MID([DividendDate],5,2) + “/” + RIGHT([DividendDate], 2) + “/” + LEFT([DividendDate],4))Bonus question:DATE(MID([DividendDate],5,2) + “/01/” + LEFT([DividendDate],4))

Question 38

Do you have a Modulus function in excel it is Mod()?

Answer 38

Yes we do have a modulus function. Instead of MOD(), Tableau has a modulo operator, in calculated fields this is the “%” character, e.g.: [Dividend] % INT[Divisor]The modulo operator finds the remainder when dividing by integers: 5%2 = 1

Question 39

Can I use calculations from Excel?

Answer 39

Yes. Once connected to the excel sheet, calculated fields are represented inside of Tableau. A change to the calculation in excel will result in updated information in the Tableau worksheet. Note: you cannot use VB or other excel functions inside of Tableau, only the output of those functions/calculations.

Question 40

Can I use functions in my database that Tableau doesn’t have?

Answer 40

Yes. You can use the various Tableau “RAWSQL” functions to return results from functions in the underlying database. Care must be taken to present the results of these functions in meaningful ways (i.e. a db function that returns a list might need to be ‘counted’ in tableau) (stored procedures may also work, as long as they return a single value for each row – aka single-valued functions)

Question 41

I have financial data. How do I calculate the profit ratio?

Answer 41

Tricky question. Their table will look like this:Account AmountGross profit 100Expenses 1049Revenue 2302COGS 8239Etc.Best way is a couple of calculated fields, e.g Profit: IF [Account] = “Gross Profit” THEN [Amount] ELSE NULL ENDRevenue: IF [Account] = “Revenue” THEN [Amount] ELSE NULL ENDEtc, then, perform your math on these calc fields: sum(Profit)/sum(Revenue)

Question 42

How do I calculate year-over-year growth without using table calcs?

Answer 42

Hint: similar to the ‘financial data’ question above.

Question 43

How do I sort on a particular dimensions’ member (i.e. sort on 2008 of “order date”)

Answer 43

I.E Do this, but without single-click sort:

Question 44

Can Tableau calculate the result of compounding interest?

Answer 44

Yes. Tableau can certainly do this with some table calculation functions:PREVIOUS_VALUE(1)*(1+AVG([Discount]))

Question 45

Can Tableau do z-tests?

Answer 45

Yes, the calculation is: (Sum(x) – WINDOW_AVG(Sum(x))) / STDEVP(x) where x is a field.Z represents the number of standard deviations between x and its population average. As a general benchmark, any Z scores above 2 are considered statistically significant.

Question 46

The Tableau Server maintenance views do not fit my needs. Can I customize them?

Answer 46

Yes. See the admin guide or knowledge base for details.

Question 47

How can I embed a view without the toolbar?

Answer 47

Add to the end of the url: ?:embed=yes&?:toolbar=no

Question 48

Can I do a multiple regression?

Answer 48

No. But Tableau does iterative regressions with ANOVA very well.

Question 49

What does the Anova table tell me?

Answer 49

analysis of variance: 1. Sources of variation (AKA “Field”) 2. Degrees of freedom 3. Sums of squares 4. Mean squares 5. F-statistic 6. P-Valuefor further information, see also: the internet.

Question 50

What is a p-Value?

Answer 50

A p-value shows the probability that sample data will adequately represent the population from which it was drawn. The larger the value, the less likely that the sample data accurately represents the population. The max. value for a P-Value is 1.0, min is zero, and a value of .05 means that there’s a 5% chance that the model does not represent the data.

Question 51

Can I use a cumulative distribution function?

Answer 51

Most likely. It depends on exactly what they are asking, but start with a histogram and set the measure to be ‘running total’. If they like that, then set it to percent of total as well (secondary table calc).

Question 52

Can I find the correlation?

Answer 52

Yes – compares two sets of data against each other and contains two separate sets of measures.

Question 53

Can I do significance testing?

Answer 53

Significance is built-in to the ‘describe trend model’ command. A possible advanced approach is to calculate your significance manually, however this requires a detailed understanding of your dataset and what significance you are testing for.

Question 54

What is scalability?

Answer 54

The ability to support multiple simultaneous actions, instead of making a single action faster.

Question 55

What is availability?

Answer 55

The ability of a solution (system) to resist component failures. Increasing availability increases cost.

Question 56

What does ‘scaling up’ mean?

Answer 56

Adding more resources (CPU, RAM, etc.) to a single server.

Question 57

What does ‘scaling out’ mean?

Answer 57

Adding more resources (CPU, RAM, etc.) by adding more servers in a “cluster.”

Question 58

What does ‘multi-process’ mean?

Answer 58

Running multiple instances of a process or service on a single server or distributed across multiple servers. This increases throughput.

Question 59

What does ‘multi-threaded’ mean?

Answer 59

Within a process, the ability to perform multiple tasks simultaneously across multiple CPUs on one machine.

Question 60

What does ‘single point of failure’ mean?

Answer 60

Within a solution (system), a component that, if it fails, will cause the solution (system) to fail as a whole or be completely inaccessible.

Question 61

How do context filters work in 9.0?

Answer 61

In 9.0, context filters create sub-queries by default. Temp tables are only created in the case of generic ODBC data sources or customized data sources (when the database isn’t set up to use sub-queries, it reverts to using temp tables).

Question 62

What are the advantages of using a subquery-based approach over using temp tables?

Answer 62

It enables join culling, considers only columns required for the viz, and applies all query optimization rewrites.

Question 63

What is the Cache Server?

Answer 63

The new Server process for the external query cache. It’s a query cache distributed and shared across the server cluster. This in-memory cache speeds user experience across many scenarios. VizQL server, backgrounder, and data server (and API server and application server to a lesser extent) make cache requests to the cache server on behalf of users or jobs. The cache is single-threaded, so if you need better performance you should run additional instances of cache server.

Question 64

Which processes use the Cache Server?

Answer 64

VizQL, backgrounder, and data server. Install the cache server on every machine running one of these processes.

Question 65

What is the default cache capacity?

Answer 65

512 MB per cache server instance. The cache is single-threaded, so if you need better performance you should run additional instances of cache server.

Question 66

What is the Repository?

Answer 66

The repository is a database that stores server data. It’s based on PostgresSQL. The server data it stores includes information on Tableau Server users, groups and group assignments, projects, permissions, datasources, extract metadata, and refresh information

Question 67

Data Server

Answer 67

The component that provides central management of Tableau Data Extracts and database connections. Allows you to upload and share data extracts, preserve database connections, and reuse calculations and field metadata.

Question 68

Why buy Server when Reader is free? What problems does Server solve?

Answer 68

Server solves problems around automation (there doesn’t have to be a person sending updates every time the data updates), creates a single version of the truth (we don’t have tons of workbooks and excel spreadsheets out there, we don’t have all these people writing their own definitions), has security features (who are you? are you allowed to see what you want to see? by having a place where the data is centrally locked down, everything’s going through Tableau Server to the customer), and gives mobile access to views (Server is browser agnostic and mobile-aware. You publish once and you can view it anywhere through any devices you want. You don’t have to have a laptop to look at your reports)

Question 69

Why is it a bad idea to take Server apart and put gateway in the DMZ and the rest of the Server behind the firewall?

Answer 69

That creates security risks. Every Tableau component ahs its own port. When you put gateway in the DMZ, you’re opening 10+ ports to the outside world on your firewall. A better practice would be to have your Proxy in front of server and keep Server all together

Question 70

What are the benefits of having a black box installation?

Answer 70

Ease of use! We have a frequent upgrade schedule, and black box installations make it easy for customers to always have the latest version of Server by taking away repetitive tasks. Our mission is to help anybody see and understand data, and our black box installation helps us do that by making it easier for people with less IT skills be able to implement Tableau, because the installation of Tableau Server is really easy!

Question 71

What’s the difference between published data sources and embedded data sources?

Answer 71

Published data sources can be used in multiple workbooks. Embedded data sources are associated with one workbook and describe the attributes required for connecting to a data source, like server name, database name, and authentication credentials. If three workbooks are connected to the same database, then you’ll have three embedded data sources listed on the Data Sources page.

Question 72

What is a black box installation?

Answer 72

A black box installation means that all the components needed to run a Tableau Server are included within a Tableau Server. Extra components (like an external load balancer) can be added later for performance, but aren’t necessary.

Question 73

What are disadvantages of a black box installation?

Answer 73

Some customers want to know specifically what’s in a Server or want to customize their Server. We’re transparent about the components and the degree of customization for a Tableau Server.

Question 74

What is Data Server?

Answer 74

The Data Server servers as a proxy between requests for data and individual data sources.The component that provides central management of Tableau Data Extracts and database connections. Allows you to upload and share data extracts, preserve database connections, and reuse calculations and field metadata.

Question 75

What are the advantages of using Data Server?

Answer 75

Data Server allows for central management of Tableau Data Extracts and database connections. You can write complex calculations and make changes to the metadata once. You only have to install drivers once on one machine. You have one source of truth and prevent data creep, since multiple workbooks can be based on the same data source. You can set up data models once. Central permission management for sensitive data sources.

Question 76

Describe the Extract API in one sentence.

Answer 76

The Extract API is a programmatic way to generate Tableau Data Extracts without manual intervention in Tableau Desktop.

Question 77

Name 4 use cases for the Extract API.

Answer 77

• Times when you want to pre-process or access and assemble data from other applications before working with it in Tableau
  
  • To connect to data sources that Tableau doesn’t have a native connector for
  • Useful for technology partners who’d like to have an “export to Tableau” button
  • Some ‘portal’ scenarios when there are workflows in which Extracts would be created on the fly instead of waiting for someone to set them up in Desktop

Question 78

What languages can you write Extract API scripts in?

Answer 78

C++, Java, and Python

Question 79

What can be done with the generated TDE?

Answer 79

Connect to it as a data source in Tableau Desktop or publish it up to Tableau Server as a published data source and go from there. In other words, the TDE can be used natively in Tableau Desktop or published to Tableau Server using the same API.

Question 80

Does the Extract API work in 32-bit or 64-bit?

Answer 80

Works in both

Question 81

What platforms is the Extract API available on?

Answer 81

It’s available to developers on Windows and Linux platforms.

Question 82

Explain what addressing and partitioning are.

Answer 82

These have to do with table calculations. Addressing indicates the direction we want our table calculation to go in, or what part of the table you are computing along. Partitioning indicates the buckets we want our calculations done on, in other words, how to group the calculation.

Question 83

Describe the Server’s process when it’s looking for something in the query cache.

Answer 83

External query cache: we save the results from previous queries for fast access by future users. Each process has an in-memory cache called the query cache. The server process first tries to look for what it needs in the query cache. If it doesn’t find it in the in-memory cache, tries to find what it needs in the Cache Server. If the result is in the Cache Server, it’s copied to the in-memory cache and returned. If it’s not in either place, the query is run on the database and the results are cached in a Cache Server as well as the in-memory cache of the process that needed it. Caches in each Cache Server are accessible by all server processes and nodes in the whole cluster.

Question 84

What’s concurrency in terms of Tableau Server?

Answer 84

For Server, concurrency is defined as the number of end users that are actively loading and interacting with visualizations at a specified response time and throughput goal. Server can typically support 10% concurrency. In our testing, we observed that Tableau Server 9.0 can support up to 927 totoal users on a 16-core single machine deployment, so 92.75 concurrent users. Scales linearly. 2 node 32 cores – 1380 total 138.04 concurrent. 3 node 48 cores 2809 total 280.93 concurrent

Question 85

What happens when you create an extract?

Answer 85

§ Tableau takes each column, creates two buffers, each 8mg a piece
§ Converts the row return data into columns
§ Takes each column, buffers it, sends it to disk
§ Then goes internal processing local to tableua to create column store, memory maps, sorting
§ When you see the update screen when creating an extract, you can see what’s going on
§ As it’s writing to the disk to create that extract, the longer that takes, the more latency you have in extract creation. That’s not usually the problem
§ Typically, problem is how long it takes for DB to spit the data out
§ When you get to higher quality databases, he’s seen 2 to 3 million rows a minute
§ Try to set the bar low w/tableau so they’re thrilled when Tableau beats it

Question 86

Why multi-node?

Answer 86

• Why multi-node?
		○ Availability
		○ Performance
		○ Scaling
		○ People in different time zones

Question 87

Which backup processes are always active?

Answer 87

• w/exception of primary every backup is up

○ Gateway is manual failover. The only passive machine in the whole cluster