Improving Automotive Battery Sales Forecast Vinod Bulusu, Haekyun Kim.

Slides:



Advertisements
Similar presentations
Decomposition Method.
Advertisements

Logistics Network Configuration
Qualitative Variables and
Chapter 5. MARKET MEASUREMENT BA L.P.Chew
11.1 Introduction to Response Surface Methodology
McGraw-Hill/Irwin Copyright © 2013 by The McGraw-Hill Companies, Inc. All rights reserved. Chapter 7: Demand Estimation and Forecasting.
Class 17: Tuesday, Nov. 9 Another example of interpreting multiple regression coefficients Steps in multiple regression analysis and example analysis Omitted.
Consistency of the Weather Nicole Baratelle, Cara Barskey, Youjin Kwon.
Session 11: Model Calibration, Validation, and Reasonableness Checks
1 Spreadsheet Modeling & Decision Analysis: A Practical Introduction to Management Science, 3e by Cliff Ragsdale.
Chapter 13 Forecasting.
Chapter 8. Organizational Demand Analysis BA L.P.Chew
1 Econometric Load Forecasting Peak and Energy Forecast 06/14/2005 Econometric Load Forecasting Peak and Energy Forecast 06/14/2005.
Slides 13b: Time-Series Models; Measuring Forecast Error
Fall, 2012 EMBA 512 Demand Forecasting Boise State University 1 Demand Forecasting.
LSS Black Belt Training Forecasting. Forecasting Models Forecasting Techniques Qualitative Models Delphi Method Jury of Executive Opinion Sales Force.
SALES Budget & sales QUOTA
Chapter 10 Target Markets: Segmentation, Evaluation, and Positioning
Demand Management and Forecasting
Michael Abbott The Impacts of Integration and Trade on Labor Markets: Methodological Challenges and Consensus Findings in the NAFTA Context.
Copyright ©2004 Global Insight, Inc. Analyzing the Home Improvement Market Market Sizing and Forecasting Mike Sweet Senior Consultant, Business Planning.
Defining the Research Problem
Target Markets: Segmentation and Evaluation
Why Normal Matters AEIC Load Research Workshop Why Normal Matters By Tim Hennessy RLW Analytics, Inc. April 12, 2005.
1 Spreadsheet Modeling & Decision Analysis: A Practical Introduction to Management Science, 3e by Cliff Ragsdale.
Using Neural Networks to Predict Claim Duration in the Presence of Right Censoring and Covariates David Speights Senior Research Statistician HNC Insurance.
DSc 3120 Generalized Modeling Techniques with Applications Part II. Forecasting.
Time Series Analysis and Forecasting
Managerial Economics Demand Estimation & Forecasting.
Operations Fall 2015 Bruce Duggan Providence University College.
Analyzing Supply Chain Performance under Different Collaborative Replenishment Strategies AIT Masters Theses Competition Wijitra Naowapadiwat Industrial.
Modeling and Forecasting Household and Person Level Control Input Data for Advance Travel Demand Modeling Presentation at 14 th TRB Planning Applications.
June 20, 2007 MBA 555 – Professor Gordon H. Dash, Jr. Determinants of Sam Adams Beer Leah Semonelli Iryna Sieczkiewicz Meghan Smith Adamson E. Streit.
Part I THE BIG PICTURE Sales Management Resources: Estimating Potentials and Forecasting Sales.
2009 Commercial GSO Demand Forecast 21 May 2009 Kevin Reyes Director, Business Development Boeing Launch Services Cover art by John.
Copyright © 2011 by The McGraw-Hill Companies, Inc. All rights reserved. McGraw-Hill/Irwin Model Building and Model Diagnostics Chapter 15.
Blended Value Accounting and Social Enterprise Success Title.
Welcome to MM305 Unit 5 Seminar Prof Greg Forecasting.
1 Where Is My Market? Mining Data to Find a Niche Commercial Lines Segmentation Workshop Lisa Sayegh Presentation to the CAS March 2003.
Market Analysis…WHY?  Gather information about your industry  Identify prospective customers and their buying habits You can not fulfill the Marketing.
INTEGRATION OF THE SALES FORCE: AN EMPIRICAL EXAMINATION Anderson, E., & Schmittlein D. C., Rand Journal of Economics, 1984 Youngsoo Kim, BADM 545 Fall.
1 1 Chapter 6 Forecasting n Quantitative Approaches to Forecasting n The Components of a Time Series n Measures of Forecast Accuracy n Using Smoothing.
Downscaling Global Climate Model Forecasts by Using Neural Networks Mark Bailey, Becca Latto, Dr. Nabin Malakar, Dr. Barry Gross, Pedro Placido The City.
Managerial Decision Modeling 6 th edition Cliff T. Ragsdale.
What we give up to do Exploratory Designs 1. Hicks Tire Wear Example data 2.
Demand Management and Forecasting Chapter 11 Portions Copyright © 2010 by The McGraw-Hill Companies, Inc. All rights reserved. McGraw-Hill/Irwin.
DEMAND FORECASTING & MARKET SEGMENTATION. Why demand forecasting?  Planning and scheduling production  Acquiring inputs  Making provision for finances.
Forecasting. Model with indicator variables The choice of a forecasting technique depends on the components identified in the time series. The techniques.
Chapter 11 – With Woodruff Modications Demand Management and Forecasting Copyright © 2010 by The McGraw-Hill Companies, Inc. All rights reserved.McGraw-Hill/Irwin.
Welcome to MM305 Unit 5 Seminar Dr. Bob Forecasting.
Welcome to MM305 Unit 5 Seminar Forecasting. What is forecasting? An attempt to predict the future using data. Generally an 8-step process 1.Why are you.
Lecture 9 Forecasting. Introduction to Forecasting * * * * * * * * o o o o o o o o Model 1Model 2 Which model performs better? There are many forecasting.
Pertemuan 11.
Chapter 5: Target Markets: Segmentation and Evaluation
Strategy and Sales Program Planning
Market Segmentation and Targeting
26134 Business Statistics Week 5 Tutorial
Correlation and Simple Linear Regression
Effective Sourcing Processes for Logistics Service Providers
Chapter 4: Seasonal Series: Forecasting and Decomposition
Global and Regional Electric Motor Sales Industry Production, Sales and Consumption Status and Prospects Professional Market Research Report.
SKU Segmentation for a Global Retailer
Perfecting Visibility
Introductory Econometrics
The Independent and Joint Effects of the Skill and Physical Bases of Relatedness in Diversification. Moshe Farjoun (1998), SMJ Weihao Li.
Quantifying the Impact of Deployment Practices on Interplant Freight Volatility Kurn Ma Manish Kumar.
Catalog Desk Impact and Opportunity Analysis
Title Evaluating the Effect of Personality Congruence between Sales Agents and Sales Managers on Performance.
1/18/2019 ST3131, Lecture 1.
EQUATION 4.1 Relationship Between One Dependent and One Independent Variable: Simple Regression Analysis.
Presentation transcript:

Improving Automotive Battery Sales Forecast Vinod Bulusu, Haekyun Kim

Agenda Company Overview Problem Statement Motivation Methodology Results and Validation Conclusion

Company Overview Thesis sponsor is a global technology and industrial leader Global leader in the after market sales of the lead-acid battery market (40% market share) Sales are through automotive retailers

Agenda Company Overview Problem Statement Motivation Methodology Results and Validation Conclusion

Problem Statement Forecasting of the sponsor company is based on historical sales, without any other variables In 2013 the demand was higher than usual, which resulted in lost sales as manufacturing could not be ramped up Based on Historical salesNot Flexible Higher demand Lost Sales

Agenda Company Overview Problem Statement Motivation Methodology Results and Validation Conclusion

Motivation -Improve sales forecast Identify whether relationship exists between sales and temperature and thus improve the sales forecast and potentially reduce lost sales Not Flexible Higher demand Lost Sales Current State Based on Historical sales

Motivation -Improve sales forecast Identify whether relationship exists between sales and temperature and thus improve the sales forecast and potentially reduce lost sales Based on variables such as temperature Not Flexible Demand Variability but NO Lost Sales Future State

Why Temperature -Based on evidence from three orthogonal sources Empirical evidence: Based on the literature review temperature impacted battery failure Preliminary (qualitative) analysis: Based on visualization of sales, we found that sales increased during the time of polar vortex Anecdotal evidence: Based on feedback from SME’s in our sponsor company

Agenda Company Overview Problem Statement Motivation Methodology Results and Validation Conclusion

Methodology 3-step process Data Collection POS Temperature Data Analysis Visualization SKU Selection City Selection Data Modeling & Validation Regression Validation Step 1: Temperature and sales data collection Step 2: Identifying appropriate SKU’s for modelling Step 3: Regression and model validation

Methodology: Step 1, Data collection Sales Data and Temperature information Weekly Point of Sales (number of units) was obtained from thesis sponsor Weekly temperature (Maximum and Minimum) information obtained from NOAA * NOAA (National Oceanic and Atmospheric Administration) POS Data Temperature (Boston)

Methodology: Step 2, Data analysis Selection of SKU Point of Sale information consisted of multiple SKU’s Appropriate SKU needed to be identified to reduce confounding with other variables such as demographics, regions, battery types etc. SKU 65 was selected based on volume of sales and geographical prevalence Volume of Sales Geographical prevalence of 65

Methodology: Step 2, Data analysis Selection of Cities Selected cities: Los Angeles, Boston, Houston, Washington D.C. and, Chicago Aggregated several zip codes in the selected cities 5 Selected cities Several zip codes in Boston

Methodology: Step 3, Modeling of SKU 65 sales with temperature Regression Modeling (Using JMP) Dependent Variables (Y-parameter): -Normalized sales as a continuous parameter Independent variables (X-parameter): -Minimum Temperature, as a continuous parameter -Maximum Temperature, as a continuous parameter -Year, as ordinal -Quarter, as ordinal 6 Models with different combinations of variables were created

Agenda Company Overview Problem Statement Motivation Methodology Results and Validation Conclusion

Model selection: Balance of Fit and Parsimony 6 Models with different combinations of variables were evaluated for fit and parsimony The selected model included quadratic and linear effect of minimum temperature and quarter ParsimonyFit Interaction of quarter and temperature Cubic and Quadratic effects R2R2 Predictability Over fit

Regression Model Maximum temperature and Year were not significant

Model validation - Is our approach valid? Validate across time (Model T) Validate across geography (Model G) Model To validate our approach we needed to validate the model over time and geography Two models were created: -Model G: Model with data from LA, Chicago and Houston (Validate the model for Boston and Washington D.C.) -Model T: Model with data from 2011, 2013 and 2014 (Validate the model for 2012)

Validating the model across time: Model T Predicted vs. actual values are compared for 2012 Trends are similar and so are absolute values

Validating the model across geography: Model G Predicted vs. actual values are compared for Boston and Washington D.C. Trends are similar, but the absolute values are significantly different Boston Washington D.C.

Validating the model across geography: Model G, cont. Predicting the % change rather than absolute value The % change is predicted well for both Boston and Washington D.C. Boston Washington D.C.

Agenda Company Overview Problem Statement Motivation Methodology Results and Validation Conclusion

Conclusion -Use minimum temperature to improve sales forecast There is a linear and quadratic relationship between temperature and battery sales The model predicts absolute sales better over time than geography Prediction of % change is similar across geography but absolute value cannot be determined by Model G -Normalizing the sales data by Vehicles in Operation (VIO) is not enough -Other factors needs to be considered: Demographics, Public transport system, Infrastructure, Initial Value The model can be further refined by -Adding age of the battery as a failure -Additional geographical data -Additional SKUs

Acknowledgements Roberto Perez-Franco (Thesis Advisor) Center for Transportation and Logistics (CTL) Thesis sponsor company