1226A: iQue on IBM BlueMix can increase your TM1 forecast accuracy · 2017-05-18 · 1226A: iQue on IBM BlueMix can increase your TM1 forecast accuracy Michael Cowie Dallas Crawford

1226A: iQue on IBM BlueMixcan increase your TM1 forecast accuracy

Michael CowieDallas Crawford

QueBIT Consulting, LLC

Mike Cowie, Director of Strategic Solutions, QueBIT

@Michael_D_Cowie

https://www.linkedin.com/in/cowiemichael

[email protected]

Dallas Crawford, Senior Sales Executive, QueBIT

https://www.linkedin.com/in/dallas-crawford-mba-446a759

[email protected]

Presenting Today

2 5/18/2017

Agenda

3 5/18/2017

• About QueBIT

• What is iQue?

• What are predictive demand planning & Euclid?

• How to get started with predictive solutions & iQue

• Q & A

What you will learn today

4 5/18/2017

1. Learn about IBM BlueMix, web services and why we created the iQueConnector

2. Learn how your forecasts can benefit from data science & predictive demand planning

3. Learn about QueBIT’s Euclid engine and predictive solutions

4. Some demonstrations to show you what’s possible

5. QueBIT’s vision for the future:

• Integrating web services and other solutions with TM1

• Leveraging advanced analytics in TM1

Reach out to QueBIT at Vision

5 5/18/2017

• On Twitter

• #askquebit or @QueBIT_TM1

• Expo hall:

• Booth #11

• More about iQue/sign up for beta: http://www.quebit.com/iQue

• Visit us on our IBM Vision 2017 page for more:

• http://quebit.com/who-we-are/resources/vision-2017/

About QueBIT

6 5/18/2017

• Focused on all things Analytics

• Analytics Consulting Services

• Project Implementations• Staff Augmentation/Managed Services

• Software Reseller and Development Company

• 16 years in business with managers on the team who have been working in analytics for 20+ years

• Operating Nationally in the United States of America

• 900+ successful implementations in finance, sales, marketing and operations (50+ new analytics implementations over the past 12 months)

• 400+ analytics customers; all industries

• 90 employees

• Development of Analytics Solutions such as Predictive Insurance Claims Subrogation, Predictive Demand Planning and Predictive Assortment Planning

• Delivering Analytics On-Premise and in The Cloud

7 5/18/2017

About QueBIT

Trusted Experts in Analytics

8 5/18/2017

• 50+ Financial Analytics Projects a year

• Delivered with IBM Planning Analytics/TM1 on Premise or in the Cloud

• Financial Consolidations and Reporting

• Revenue and Expense Planning

• Operations and Workforce Planning

• Capital Planning

• Profitability Analytics

• Etc.

Financial & Operational Performance Management

9 5/18/2017

• CloudWORQ: Pre-built TM1 solutions designed to integrate with TM1 Cloud, based on proven QueBIT TM1 models, such as expense, workforce & sales planning

• ControlWORQ: Excel & Web-based Financial Consolidations with TM1 (journals, eliminations, etc)

• ReportWORQ: Report bursting and distribution solution for TM1

• WebWORQ: Web application development for TM1, including dashboard and visualization capabilities

10 5/18/2017

QueBIT FrameWORQSolutions

• Dozens of Predictive and Prescriptive Analytics project a year

• Dramatic ROI

• Solution Focus • Galileo – Predictive Demand Planning

• Orion – Predictive Assortment Planning

• Athens – Predictive Claims Subrogation

• Predictive Price Optimization

• IOT Analytics

• Predictive Maintenance

• Predictive Customer Intelligence

• Scheduling Optimization

• Data Scientist for Rent!

Advanced Analytics

11 5/18/2017

• Cognos Analytics Reporting & Analysis

• Dashboarding – KPIs/Scorecards

• Production Reporting

• Sales

• Supply Chain

• Operations

• 75% of clients now claim they are moving to Mobile BI

• Data Warehousing and Big Data

• QueBIT can help with Data Strategy

BI and Data Management

12 5/18/2017

What is iQue

13 5/18/2017

TM1

Flat Files

ODBC & ODBO

TM1 Views/Subsets

Package Connector

Execute Command

Java Extensions

What can we see & do from a TM1 Server?

14 5/18/2017

Cloud & On Prem

On Prem Only

15 5/18/2017

But… what about?

• Non-ODBC data sources

• External applications & services: Data

• Weather, Markets, Twitter, etc.

• External applications & services

• Watson, Messaging, Monitoring

• Anything found in IBM BlueMix & similar platforms

• https://console.ng.bluemix.net/catalog/?category=services

What is iQue

16 5/18/2017

TM

1 R

ES

T A

PI

3rd

Pa

rty S

DK

s, A

PIs

, e

tc.

• Wizards to help define connections & mappings between TM1 instances and 3rd party services

• Connectors that can be used to execute those saved connection and mapping definitions

• Current support for: • QueBIT’s time series forecasting engine (Euclid)

• SMS messaging

• Future support: Customer requests/demand (e.g., Watson & other BlueMixservices)

What is iQue

17 5/18/2017

Demo

Texting while TM1’ing

18 5/18/2017

What is Predictive Demand Planning?

19 5/18/2017

• Collaboration of data science (predictive modeling/forecasting), sales, inventory management and finance

• Data science driven forecasts that maximize accuracy by leveraging causal data

• Forecast Value Add metrics/alerts that identify potential issues quickly/efficiently

• Collaborative environment where sales, inventory management and finance can work together to optimize inventory levels, model P&L impacts and perform “what if” scenario testing

What is Predictive Demand Planning?

20 5/18/2017

• Traditional Demand Planning Solutions:• Are based on naïve seasonality and trend forecasting techniques and

do not include causal data such as weather, pricing, promotions, macro-economics effects

• Struggle with new items

• Struggle with low velocity items

• Present difficulties with changing drivers and structures within database

• Are not interactive and do not allow for “what if” scenario testing

• Are not truly collaborative: sales, IM and finance

Why would you need predictive demand planning?

21 5/18/2017

QueBIT’s Unique Approach to Forecasting

22 5/18/2017

QueBIT’s Unique Approach to Forecasting

23 5/18/2017

Product sales by month – how do

predictive and traditional tools

compare?

QueBIT’s Unique Approach to Forecasting

24 5/18/2017

Seasonality can be extracted by

removing the overall trend

The trend is decidedly non-linear and

drops off dramatically in 2014

QueBIT’s Unique Approach to Forecasting

25 5/18/2017

QueBIT’s Unique Approach to Forecasting

26 5/18/2017

But other forecasting tools assume

linearity and would create a trend

below (and then layer on seasonality)

The sales trend is highly correlated

with overall marketing spend –

therefore, we want to predict the past

(and future) demand trend using

marketing spend (and other

predictors)

QueBIT’s Unique Approach to Forecasting

27 5/18/2017

QueBIT’s Unique Approach to Forecasting

28 5/18/2017

Our approach generates a more

realistic forecast by incorporating

causal data (such as marketing

spend)

What is Euclid?

29 5/18/2017

• Euclid is QueBIT’s native Java predictive modeling engine

• Extremely high performance (multi-threaded)

• Contains proven, unique QueBIT intellectual property (based on deep knowledge of data science)

• REST API that can be leveraged by TM1 using iQue

What is Euclid?

30 5/18/2017

Algorithms supported:

• Forecasting:• Seasonal Forecasting Model with Causal Predictors• ARIMA

• Clustering/Segmentation:• KMeans (with automatic # of clusters optimization)• KMedians

• Predictive Modeling:• Multiple Linear Regression• RandomForest• C5 Decision Trees• Logistic Regression• Deep Neural Networks

• Text Mining:• Latent Dirichlet Analysis (aka Topic Modeling)

What is Euclid?

31 5/18/2017

Availability:

• On-premise

• Hosted pay-as-you-go service

What is Euclid?

32 5/18/2017

Demo

Call the Euclid Predictive Demand Planning Engine from TM1

33 5/18/2017

• A set of dashboard pages created in Planning Analytics Workspace (PAW)

• How we define mappings from existing TM1 models to Euclid

• Execution of a pre-defined iQue package that calls QueBIT’s Euclid predictive demand planning engine:

• Approximately 100,000 data points consisting of historical weekly data (going back 2 year) and predictors (unit price, ad & digital spend) extracted from TM1 and sent to Euclid

• Euclid returned predictive forecast for the next 52 weeks along with a lower and upper confidence interval for the same period, which was then loaded into TM1

• Increased digital ad spend by 400%, re-ran Euclid prediction to see impact

What did you just see?

34 5/18/2017

Questions?

35 5/18/2017

• On Twitter

• #askquebit or @QueBIT_TM1

• Expo hall:

• Booth #11

• More about iQue/sign up for beta: http://www.quebit.com/iQue

• Visit us on our IBM Vision 2017 page for more:

• http://quebit.com/who-we-are/resources/vision-2017/

Thank You!

Other Sessions with QueBIT• 1148 (Wed @ 11:15am): Leveraging Planning Analytics & CDM to streamline Financial Reporting and improve

Budgeting at Stein Mart

• 1095 (Wed @ 3:00pm): Prescriptive Analytics & CPLEX Decision Optimization and TM1 Dashboard: Combined

Planning Power at JLG Industries

• 1091 (Thurs @ 9:15am): Auditable Financial System for Government Contracting at Accenture Federal Serv

• 1220 (Thurs @ 10:45am): Planning Analytics: Working Alongside Hyperion for Operational Analytics at NiSource ices

• 1093 (Thurs @ 10:45am): Leveraging TM1 for Merchandise Planning at Tractor Supply Company

Be sure to stop by and visit us at Booth #11 and…

Visit us on our IBM Vision 2017 page for more:

http://quebit.com/who-we-are/resources/vision-2017/