This document discusses how Power BI can be used for both corporate business intelligence (BI) and data science (DS) as well as personal BI and DS. For corporate use, Power BI can create centralized, reusable and scalable solutions for BI and traditional research-based DS. For personal use, Power BI allows users to easily create their own BI solutions and explore DS by cleaning, analyzing and building models on data without extensive IT support. The document provides an example of how an individual could use Power BI to perform demand forecasting for a public biking system by obtaining data, ensuring quality, exploring patterns, developing features, forming hypotheses and applying machine learning.
Advanced Machine Learning for Business Professionals
Personal bi to personal data science
1. Personal BI to
Personal Data Science
Mulkens Jan Hermans Kimberly
Microsoft BI Consultant Data scientist
Twitter: @JanMulkens
Blog: www.janmulkens.be
3. Jan Mulkens
Microsoft BI Consultant
Ordina Belgium
“I love the Power BI experience from desktop to mobile.
Put on top of that the community and the continuous stream
of high value releases and it’s clear why Power BI is the future.”
4. Kimberly Hermans
My favourite thing about Power BI is its ease to work with.
Within hours I know what my data looks like, what story it tells
and how I should start my modelling phase.
Data Scientist & CRM consultant
Ordina Belgium
5. Agenda
• Introduction
• Corporate BI & DS
• Goals
• Solution
• Personal BI & DS
• Goals
• Solution
• Applying Power BI “With Power BI, it’s simple. Everything becomes possible.”
Philip Dean, Tees and Hartlepool National Health Services Trust
6. Introduction
Power BI can not only be used to democratize BI across
the enterprise, but also to democratize Data Science
across the enterprise.
17. CorporateDataScience
Up to
15hrs / week
Up to
15hrs / week
Up to
15hrs / week
Full work week
Up to 4 hrs
/ week
Source: Generalized from O’Reilly’s “2015 Data Science Salary Survey” (sep 2015)
ETL Data Cleaning Machine LearningExploratory Data Analysis
25. Demand forecasting
• Public biking system Washinton
• +/- 350 stations
• Peaks in demand & supply
• Goal of project:
optimize planning of relocating bikes
64. PersonalDataScience
2) Get Data
5) Feature engineering
4) Exploratory Data Analysis
3) Enforce Data Quality
1) Problem statement
6) Form hypotheses
69. PersonalDataScience
2) Get Data
5) Feature engineering
4) Exploratory Data Analysis
3) Enforce Data Quality
1) Problem statement
6) Form hypotheses
7) Machine Learning
Kimberly:
The goal of personal bi & ds is to give everyone an as flexible as possible access to the data they need to be able to do their job, create a report and even make strategic decisions.
Or in other words we have to be business driven instead of it-driven. The bussiness should be able to get the data they need instead of creating change request for eeeevvveeeryyyy report they need.
Jan:
When we become more business driven, this also means that the business won’t have to wait untill IT finds a minute of spare time to handle a minor request and thus indirectly... everything will be delivered so much faster.
Who here is familiar with this situation at the moment or has been in the situation that business is waiting for a minor change, while IT has no time due to major incidents or projects?
So you are familiar with this situation, then you can imagine that by creating a more personal environment that there will be less administration and to be honest, less irritation and by extension ‘less’ governance.
Kimberly:
We will have less standardized reports and more freedom to create something that suites our own personal needs.