The journey with Tableau begins

Tableau is a kickass Data Visualisation software which is used by Small and Medium Businesses as well as large corporations like Google, Uber, etc.

Let’s start our journey of learning Tableau:

Big Idea of Tableau –

Pills are of two colors:

  • Blue – These represent Dimensions – They are text or range.
    • Dimensions – They split up the view by unique values
  • Green – These represent Measures which are numeric.
    • Aggregated to the level specified by dimensions.

Raw level of granuarlity

Understanding the data – what is it, type of data, repeated values

what makes it unique or difference from other rows as that helps in creating calculations.

throw all dimensions on canvas that’s row granularit ylevel.

Aggregation like sum of sales is going up while granularity is going down the dimension – new Dimension to convas – you are increasing granularity so the more the dimensions you have, higher and deeper the granularity would be.

Dimension determine the visualisation Level of detail. The visualisation Level of Detail becomes less aggregated or more granular as more dimensions are added.

Knowing analytics lquestions – focus on N, A, V – identifying dimensions and measures – are Qat a time.

What are you trying to achieve.

Form frameworks and run through them for added accuracy and saving a lot of time.

Taking diemsnsions to shelves.

Panes -= no voz LOD level.

Drop here in the big window – whatever you see if your visualisation LOD

LOD expressions allows you to do is to define the calculation of LOD

independent building of visualisation is visualisation of LOD

LOD – Level of Detail is an important concept in Tableau.

LODs – changes the level of detail we look at

{Keyword [dim1],[dim2]: aggregate expressions} – SUM, MIN, MAX, AVG

where aggregate expressions denotes the list of dimension to include and are greater than 0 dimension used.


FIXED level of detail – can be both measure and dimesnion. These are not ignored by most filters in the view. Gets predictable filter.

{FIXED [category]:sum([Sales])}

for every category, compute sum of sales}

{fixed:sum([sales]} – compute the total um of sales.

do calculations regardless of totally agreegated

chart at country level – chart other

it is independent

deterministic can count on giving same result again depemds on the value being returned. Drag a measure to Dimension. Use as Dimensoin and measure – apply as filter.

bunch of customers – countd(customer name)

show sum of sales year over year by customer cohorts base don Customer acquisition date don’t have it qst order date of each customer name –

use the query as noun and preposition

do i have all the dimensions

need subset of data or all data

dimension need to be assigned at row level

but derived at a more aggregate level

do I need to assign anything at row level

tag each row with CAD – aggregation levle is customer

derive CAD for each customer

create LOD Expression – create row level calculation – {FIXED [customername]:(MIN[orderdatre])}

INCLUDE – the normal granuarity. They are always a measure.

Going to deeper levle than the visualisation level.

coutnry level – state level – if this is desired

include a dimension that is not part of visualisation LOD

Calculations can’t do this.

show by country, distinct count of all products and compare it to distrinct count of all unprofitable products. do not show individual products.

Country – profitable products – v/s distinct count of product names in every single country.

show now only non profitable products

measure aggregation be more granular than the visuaisation level – yes

include (product name) :iff(sum(profit<)), count distinct product name)

country v/s distinct products which are unprofitable.

EXCLUDE – going upper in the granuarity from the present granuarity in the visualisation. They are always a measure.

Questions: should you add all calculations in the data model?

Table calculations – in the shelf – you add these without level of detail. it is easy to create a lot of calculations. You can overwhelm yourself and others.

Apply best practice. How to apply LOD?

Organise it nicely. Name in meaningful wayl.

Paste in comment.

Organise LODs into folders.

How to reconcile the calculation output from LODs and Calculations?

  • Putting two views side by side.
  • Create multiple views and drop them side by side in a dashboard to reconcile.
  • Calculation v/s Desired Result.

Filter – how it operates – they have their own order of functioning – priority sequence of filters is what it is called. what happens before?

normal dimensions filter affects only calculation

table calculation filter – lookup(min(category)

calc) – only sheet level

LOD – beyond sheet level

LOD area part of query to underlying –

nested select so depeendent on DBMS -nearily impacted by join

try using aggregated extracts

add to context if not to be ignored – FIXED

relative refering absolute

table calculation can be more flexible

All LODs – fixed sales exclude sales include sales

filter order of operations

Table calculations

text table / crosss tab

use calculate assistance – add sub totals

select specific dimensions

change visualisation 0 as you want

numbers will never break

going into framework : mmoving average on weekly basis

look up previous values – table calculations is only option

syntax for moving average

current row and looking into future

dual axis – synchronise – previous weeks

window avg(sales -3.1) – 1 is upcoming weeks, 3 is window

table calcul;ation – average sales cross regions by category

show region and categories

having framework – calculations – calculate average of across region

window – average (sales) do it for region )

table calculation – build a framework

Leave a Reply

Your email address will not be published. Required fields are marked *