Which Chart do I Choose?

(In Power BI)

Data Analysis is a process of finding insights from information. Data normally be in a raw or unstructured format which needs to be formatted to get Flat structure which is used to get the insights.

Data----------Information----------Insights Normally Data comes under the category of people, who day in and day out gather the data from various places of the organization. Data analyst then work on the Data and converts it in to information on which they can do the analysis and get insights. Data Analysis Life Cycle: IDENTIFY----COLLECT----CLEAN----ANALYZE----INTERPRET First you should understand what are the data points you need let's see you are planning to do analysis on "Sales planning" where you need to know how your sales is at present, what is the growth rate, upcoming Expense planning..etc. Next collect all the data related to it. Now it's time to clean the data. You need to handle the structure, missing data, duplicate data, invalid data to make it in a way which can be used. In Analyze state you will analyze the structured data in different ways slice and dice the data to understand it. In the last stage you will write all your observations, analysis and insights and present it to the user. Types of Analysis:

1. Descriptive Analysis: Is where you analyze what is going on in the business. Answers "What" portion of the problem. If you take the example of Sales data this gives the analysis about what are our current sales, growth rate, sales per timestamp, are we in profit or loss..etc.

2. Diagnostic Analysis: Investigates "Why" part of the problem. Analyzes why the business is running the way it is now. In sales perspective it gives the analysis on why is the sales high or why are we in losses during a certain period, why are the sales not consistent..etc.

3. Predictive Analysis:

   Shows what will happen in future based on the past trends. Based on how the sales trend from the past quarter ,it sets a trend for sales for the upcoming quarter.

4. Prescriptive Analysis: Makes conclusion based on all the previous analysis and advice what to do next. Gives information like the right time to launch new product, Start giving offers..etc.

For a data analyst, it is crucial to understand which chart is best suited for each type of visualization.Before you start creating a chart, first decide what kind of analysis you are working on. There are three key questions you should ask yourself.

Let’s explore these questions and how to answer them to proceed effectively.

Question 1: What type of data are you working with? Is it Geospatial, Textual,Hierarchical, Financial,Categorical ,Time-series..etc. Understand the data you are working with.

Question 2: What do u want to communicate?

Comparison: 

Used to compare values over time or across categories.

Common Visuals: Column/Bar Chart Clustered Column/Bar Chart Data Table Heat map Line Charts Area Charts

Radar Charts

Composition: 

Used to break down the component parts of a whole. Common Visuals: Stacked Bar/Column Chart Pie/Donut Chart Stacked Area Waterfall Chart Funnel Chart Tree map Sunburst

Relationship: 

Used to show correlation between multiple variables. Common Visuals: Scatter Plot, Bubble Chart Data Table Heat Map Correlation matrix

Distribution: Used to show the frequency of values with in a series. Common Visuals: HistogramDensity, plotScatter Data Table Heat Map Box and Whisker

Question 3: Who is the end user and what do they need? Analyst, Manager, General public, Executive? The Analyst:

Likes to see details and understand exactly what’s happening at the granular level. Tables or Combo Charts, granular details to support root cause analysis.

Manager: Wants summarized data with clear and actionable insights to help operate the business. Common charts and graphs.Some details but only when it supports a specific insight.

Executive: Needs high level,Crystal clear KPIs to track business health and toppling performance. KPI cards or simple charts.Minimal details unless it adds critical context to KPIs Dashboard design Framework: Dashboards are analytics tools designed to consolidate data from multiple sources, track key metrics at a glance, and facilitate data driven story telling and decision making.

Framework: 1.Define the purpose. 2.Choose right metrics 3.Present the data effectively 4.Eliminate clutter and noise 5.Use layout to focus attention 6.Tell a clear story

A well designed dashboard should serve a distinct purpose for a distinct audience, Use clear and effective metrics and visuals and provide a simple and intuitive user experience.

Key questions to consider: 1. Who are the end users of your dashboard? 2. What are their key business goals and objectives? 3. what are the most important questions they need answers to?

4.How can I present the information as clearly as possible?

Building and Formatting Charts: The build menu allows you to change the visual type, Auto suggest visuals and add data to customize chart components. This is Contextual menu which means you get the options as per the selected visual.

You can build visuals by choosing the chart type and adding the data or Drag the fields from data pane on to the canvasFormat pane let’s you format the visuals by changing the background, title. Color etc. Enable "On-Object formatting" by double clicking the chart object (Right click-Format) which allows to select edit individual chart elements. This is available for only few visuals like Bar, Column, Line, Area, Combo & Scatter.

Filtering Options: 3 types of filters available in the report view. 1.Visual Level Filter—Applies to a specific visual. 2. Page Level Filter—Applies to all visuals on a given report  page. 3.Filters on All pages—Applies to all visuals across  all the report pages. These are the few things which you need to keep in mind before you make any visualizations. Understand the data very well and follow the steps provided.