Power BI workshop (36 hours)

Date :- 24-11-2024

In data modelling there are 2 types of table :-

1) Dimension Table  We store data which describes my business entity. For
example :- If someone asks you to collect data and store it, for banking you will
create all banking related table , like customers table, branch table, banking table
etc.
Whichever table that defines your business entity is dimension table. In this table
the column which identifies each record with unique id is called as primary key.
(Column)

2) Fact Table :- In the fact table we store data related to any transaction or event /
or facts and figures. For example :- in banking system one entity is customer 2 nd
entity is banking 3rd record is transaction table in that you have transaction id , what
type of transaction, how much was debited, what is the balance etc. These types of
data are stored in the fact table. In the fact table you have both the keys. Primary
key will only come one time and foreign key will come multiple time.

In transaction Table  All have unique ID column (primary Key)

Transaction ID Primary Key

Customer ID Foreign Key (since we
Employee ID are using this from
Dimension table) can
have multiple entries.
Type of transaction
Branch
Amount

Make power pivot for many things like 2-3 things from each file
You have to go to power pivot and then manage , here you are going to make your
data ready, on excel we shall create reports but in power pivot manage in
following :-

1) Go home  Get external data  From other sources  excel file  please press
use first row as column headers (it is important to tick the check box) . Making this
is like green room for data you first do your work in power pivot.

2) The file opened is not excel but a data model, entire section is divided into 2
parts, above is -------, below section is calculation section
3) On home tab extreme right corner  Diagram view  Make relationship (1 too
many and many too one relationship).
4) Make one too many relationships with cookie type and product. Try to create the
relationship of one too many in power bi and power pivot.

5) Go to data view. Our requirement is like to create total revenue, ii) Profit, iii) Profit
Margin, iv) Total revenue by country, v) Total quantity by customer, vi) Total margin
by product. These are the different reports we want to create from Cookie crunch
data. Here we can use

6) First created a new column for revenue in Orders, Use orders * related function
and your values will be calculated here. This type of formula what we are using is
called DAX function. Till now everything is happening in data model and nothing in
the main excel sheet.

7) The single final answer of all the values from the columns is called measure. Total
revenue and profit is calculated by formula – Total revenue:= sum(the formula) and
enter. This is done is working column of the Power pivot.

8) To go to excel (switch to excel) just go to top left corner and press the dark excel
sheet. Then insert pivot table  from data model (use data from data model and
create a pivot table in A1 cell). A blank layout with 3 tables appears.

9) Check in settings to do side by side. Highly recommended change the name of

pivot table. When creating a pivot table please make sure you make it horizontally
and not vertically. Leave some space and make pivot table.

Power BI Process:-

1) Get data Clean data Modelling  Visualization Analyze  Manage

2) Five categories of analytics :- i) Descriptive – a large data set is analyzed to get

an outcome of that data in the last 5 years, will help us in answering the WHAT
question , ii) Diagnostic – Will help us in answering the WHY question. Iii) Predictive
analysis – What will happen in future (forecasting). iv) Prescriptive Analysis :- 62
lakhs was at the end of 5 years and predictive analysis has given me 80 lakhs so
prescriptive analytics will tell me how will I reach from 62 lakhs to 80 lakhs ( it will
answer the question of how) v) Cognitive analytics :- Self learning feedback loop
( self-learning  Results  Feed back  We learnt from our feedback).
3) Roles in Data :- i) Business analyst :- a person who knows everything about the
project, he sits with the stakeholders, what are the deliverables what can be done
what have we promised, how much grace period we have etc, and then give it to
Data analyst. ii) Data analyst :- Who is well versed with cleaning data, modelling,
visualization , analyse, managing the data set, managing refreshers, etc . iii) Data
engineer :- A person who will create a data for you. He will make the table
structured, and make a table and make a data set, from different sources to one
place. From different platforms like sql, ms access and other place and keep it in
one place. Iv) Data scientist :- very well versed with coding , but doesn’t know how
to convert to visual , doesn’t know how to make relationships in data, he can do
advance analytics but reports and visuals he doesn’t know. V) Database
administrator :- Data engineer only looks for data from various data to one place,
but administrators work is platform management, network, the platform is working
or not.

OLTP Database :- Online transaction processing Database.

when we extract data from SQL server (where lots of data in form of tables is made)
we perform ETL Process  1) Extract the data , 2 ) Cleaning (transformation)
process , 3) Loading - Data warehouse (mini server) where after cleaning and all we
dump our report.

Crystal report , SSRS, Cliq View (report creating software’s).

Power BI :- On premise ETL and reporting Tool , it is a combination of power query

editor  the whole ETL process is power query editor + Power BI desktop  Data
model, Data visualization and data reports, all these are in form of a file. You can
also store it in cloud. It’s a collection of software, service and Power BI mobile
applications. (apps). Collection of all 3 things.

Only in Power BI desktop you can pull data, Model data, clean data only in the
desktop version not in cloud.

Very first work for data analysis is :- 1) Data preparation, there are different tables
are available in different platforms (like text file, excel, word file, SQL server etc).
keep in mind the requirement list, which will be different and yours will be different,
this is completely based on the requirement list of the client.

Process :- Original data  Power BI desktop  Clean your data  Model your data
 Visualise your data  And report your data.

Power BI cloud is for sharing and collaboration.

Building blocks of Power BI :- 1) Visual , 2) Dataset , 3) Report, 4) Dashboard , 5)

Tiles.

Power BI SQL mostly uses Transact SQL.

Always take table and not sheets. Whenever data in table form is available click
data from table or else take sheets when uploading a data. It is because in table the
nos will constantly get updated where as in sheets the data will be static.
DAY 2 -
1) Get data

In power BI if you want to create a relationship yourself and take help from power BI
you can uncheck 2 boxes, from the options tab.

When taking a data from direct query mode in power BI desktop it will not show, it
will only show in Power BI transform Data.

* If you have dynamic tables every 5 mins update then go for uploading direct query
mode from SQL server, and if your data is static and not being updated every 5
mins go for imported mode (also upto 1 GB).

* When using import mode it is flexible you can make relationships and make
transformations, but if you are using direct query data is only available on server
which is a relational database, there you cannot disrupt the database.

* Any direct query table when connected with import table it will always be dual
connection.

* When taking data direct from folder, make sure that sequence of
columns are same before loading the data.

* MongoDB, Assure Cosmos DB. Are databases from NoSQL databases. (here
data is not available in rows and columns.) This database will be used in social
media, since it does not have structured database.

First Name (Given) Null Last Name

Null Null Last Name
First Name Middle Name Last Name

Here The data is unstructured (NoSQL) , if NoSQL you don’t know the data engineer
will give you file in JSON file. Here go to Azzure DB and get file and take data and
load it.

* 3rd party application and you can paste url and paste it in the get data  More 
Online services  select it, put path of application and load data.
* If you want to take data from lets say money control or something you copy url
got to get data Other  Web  Enter the URL and power BI will go and check.
(you can get data from websites)

* Fix refresh problems by, Transform data  Tools  Diagnose Step  Refresh 
Stop diagnostics
* The language which is used in power query editor where you remove columns and
so many other things and then send to advanced editor, that is called M language.

* Query folding is a concept and a feature to understand.

STAGE II

3 things we will learn cleaning, transforming and ____________ .

Use power query editor to clean up and shape data.

* You have to do pivot data to unpivot. i.e sometimes we have structured data we
have to unstructured it, i.e columns to rows and rows are changed to columns.
* When unpivoting the data the two columns added to the data set are values and
attributes.

Profiling data and examining Structures

We have to find anomalies. Check column quality (Column profiling).

Transform data  View and click on column quality , here it is showing % of valid
data, errors and empty column.

2nd part is column distribution,

Column profiling :- tells us minute details of the columns like max and min avg etc.

When combining tables with rows by adding rows its called as appending

When we want to combine columns from sheets we merge

Combining datas can be possible in 2 ways append query and merge

query.

Left outer means :- All the record from table one and the matching record from table
two (merger)
Right outer means : when merge column will have the records which have values
and the remaining will be ignored
Data Modeling:

One of the important and most tedious task for a Data analyst. By doing this well we
will be able to make accurate reports, faster exploration, simple aggregations and it
will be easier to maintain.
Always follow Star Schemas i.e structure of the data model.

Fact table should come in middle and all dimensions table should be around it

Cross filter : Allows data to flow both ways, where one filter will be put in both the
tables even when you do it in one table.

DAX : Data analysis expression (is a programming language to perform analysis in

Power BI model)
We can use dax for creating Measures, calculating table, calculate column.

Cardinality : Uniqueness of data values in a column.

Dimension : Store details about business entities ,,

Hierarchy : Organize data such that one element is ranged over other data.

ROLE PLAYING DIMENSION

A dimension that can filter related facts differently. Example 3 tables inside
date table,, ship date ,due date and order date,, one date table can alternate that
fact table as it has 3 relations. But only one relationship can be active at a time.

Date 05-01-2025

Microsoft has made language (code) like in excel we do sum or if or some other
formulas, in Power BI it is called DAX. Go to Dax guide in google to find all the
functions and what are its formulas.

Important category :- Time intelligence category, iterator functions (normal

aggregator functions but ends with X). (ittretator function has repetition, level and
row filters) . In PL 300 they will not ask us how to write the functions they will just
ask what is the syntax that’s all.

Dax is used :- 1) To calculate table, column and measure.

Difference in calculate Column and measure :- in power BI when you do sum of all
values or calculate the sum total of a range, and the answer in one particular cell is
given, that calculation is called a measure. Basically it’s a summarization of data in
a single cell. Measure is stored in RAM and not in the file you are working on.
Measures are always created in Facts table.

To create measures :- Click on enter data  Click on name table  And simply press
load, a blank table is created, now this is the table where we are going to store all
measures.

In facts internt sales table we want to total sales amount, now we want to add it, so
we will create a measure, right click on all measures  Click new measure  there
automatically measure = will come so there we rename it to Total sales= (formula
bars comes with a default name you rename it) you press sum and then sales
amount  press tab and close the table with bracket and press enter.
The result of measure is not seen on table view but only can be seen in the visual,
We have to take some visual and put it in the sheet then only measure result can be
seen. If measure has to be seen you press card  then click on total sales amount
in visual you can see the amount i.e the answer of measure.

Calculated columns :- It is an extension on table, it is not directly there in power BI ,

we have to create it.
Calculated column can be created in any table, unlike measure which can be made
in facts table, we can add any calculated column in any table where ever we want, if
we want in fact table we create there or else in anyother sheet we create there. You
right click in facts table  new column  a formula bar is activated by default name
is given you can change the name  then you calculate there and press enter 
Calculated table is physically stored in table.

It is always better to have a measure instead of calculated columns.

Without a physical column I want to create a measure, so we use iterator function. It

is helpful cause you make a function where you are creating a repeater function
without creating a new column and directly doing sum and getting the final answer.

Columns vs. measures :- CC (calculate columns) creates new value for each and
every row in the column 2) CC is stored in power BI PBX. file only i.e it will increase
fill size ,
Measures :- Created on demand, measures are calculated based on filters so it will
change according to the filters, if we apply filter answer will change and if we
unfilter it will change

For power BI Understanding context is very important, it is an environment, i.e

reference to the above given topic,

Context two types :- 1) filter context and 2) Row context (which is equivalent to your
calculate column).

Iterator function :- is always SUMX, IFX or anything with X function.

Learn calculate function it is a very important function to learn as a data

analyst, which is the most important DAX function. (it is a function method
which creates a method and ignores filter context), for example :- if you want red
colour sales and no matter what filter you apply, when you create a new measure
using calculate function no matter what the filter you apply it will always have red
colour sales only.

Straight line relationship made in relationship mode :- is called active relationship

Dotted line relations ship :- is called inactive relationship.

Use relationship function helps in not going to physically making relations made
active or inactive, you can directly make measures with inactive relationships also.
There is a concept of semi-additive measure creation also in power BI

Semi additive means total karna hai but not full. (not full addition). (example you
want last date total only of sales and not the other days total). To create semi
additive function you have 2 functions one is first date and last date function can be
used to create semi additive measures.
Time intelligence :- This time last year, and many more functions.

Performance Optimization :- Variable is a temp location above ram where we

store some data.

A good practice of creating a power BI report is go with the theme, like company
theme etc.