Power of Data Visualization by Power BI. FB Page: ข ช างจ บข อม ล

Transcription

1 Power of Data Visualization by Power BI

2 Profile Rapheephan (Pear) Laochamroonvorapongse Petroleum Engineer, PTTEP B.E. in Petroleum Engineering Chulalongkorn U. (1st Class Honor) M.S. in Petroleum Engineering at The University of Texas at Austin Research Area: Data Analytics, Low Dimension Modeling, Artificial Intelligence Application Admin Page ข ช างจ บข อม ล

3 What is Power BI? What is Power BI? Power BI is a business analytics solution that lets you visualize your data and share insights across your organization, or embed them in your app or website Gartner Magic Quadrant for Analytics & BI Platform Source:powerbi.microsoft.com

4 Why Power BI? Connect to various source data and analyze million rows of data Define complex calculations using Data Analysis Expression (DAX) Visualize data with interactive report and dashboard Collaborate and share report or dashboard to team inside or outside organization More important, it is FREE!!!

5 Power BI Pricing Power BI Desktop Power BI Pro Power BI Premium Power BI Embedded Free Data Analysis and report tool 9.9 USD/monthuser Collaborate on the shared data Audit and govern how data is accessed Cost depends on number of users and capacity Large scale deployment Minimum 750 USD/month Visual analytics embeded in your applications

6 How to Install Power BI? 1. Click Download Free to start Power BI Desktop 2. Click Sign up and register with school or office 3. Open Power BI program to start program Remark: Sign in the account in order to share report to Power BI service and import new visuals from marketplace

7 Power BI Overview Report Data Relationships Three key steps of Power BI Data Relationships Report

8 Step 1: Data > Get Data Power BI can connect to various types of source data Files or Folder (csv, txt, Excel) Database (SQL, Access, Oracle, IBM, Azure, etc.) Online Services (Sharepoint, Facebook, Github, Power BI Service, etc.) Others (R script, *Python script, Spark, Hadoop, etc.)

9 Let s explore condo data

10 Step 1: Data > Get Data 1. Check worksheet that you would like to upload 2. Click Edit to go to query editor

11 Step 1: Data > Query Editor Roles of Query Editor 1. Data cleaning and transformation 2. Add calculated columns Table Name Function Bar Query List Transformation Steps

12 Step 1: Data > Define Data Category 1. Define Data Category: Data Tab > Select Data Column > Modeling > Data Type Date Text Whole Number Decimal Number True False 2. Recheck Data Category in Query Editor

13 All data are ready to cook!

14 Step 2: Data Model Manage Relationship between DATA TABLE and LOOKUP TABLE DATA TABLE: Dynamic Table (transaction, sale, return, etc.) LOOKUP TABLE: Descriptive Table (country code, product category, etc.) Good News: Power BI will automatically detect data relationship This is Data Model This is not Data Model

15 Step 2: Data Model Foreign Keys in Data Table Primary Keys in Lookup Table Avoid using Both Direction

16 All data are mixed together!

17 Step 3: Data Visualization 4. Filters (Visual, Page, Report Levels) 3. Field, Format, Analytics Panes 1. Visuals (Bar, Line, Scatter, Map,Filter, Matrix, etc.) 2. Field List (Table, Columns, Measures) Report Page (Add, Duplicate, and Delete)

18 Step 3: Data Visualization Time Filter KPI Tree Map Card Map Bar Chart

19 Step 3: Data Visualization Forecast Select Trendline from Analytics Pane Line Chart: Continuous plot Scatter Chart

20 Data are ready to be served!

21 Let s continue with condo data

22 Example of Condo Report

23 Example of Condo Report

24 Example of Condo Report

25 Example of Condo Report

26 Example of Condo Forecast

27 Python Script in Power BI (New) import seaborn as sns import matplotlib.pyplot as plt sns.distplot(dataset.sale_price_sqm, kde=false, bins=40) plt.show() g=sns.pairgrid(dataset) g=g.map(plt.scatter) plt.show() sns.lmplot(x="dist_paragon",y="sale_price_sqm", data=dataset) plt.show()

28 R Script in Power BI library(ggplot2) ggplot(dataset, aes(x=sale_price_sqm)) +geom_histogram(colour="black", fill="blue",bin=40) pairs(~sale_price_sqm+year_built+dist_paragon,data=dataset, col="blue") ggplot(dataset,aes(dist_paragon,sale_price_sqm))+ geom_point(aes(color="red",alpha=0.3))+ geom_smooth(method="lm")

29 Example of Condo Report