The Forecaster’s Imperative | September 19, 2018 Forecasting and Credibility The Forecaster’s Imperative | September 19, 2018
What is the most important feature of a forecast?
The Imperative and it’s Challenges The Imperative is Credibility The Cassandra Curse The Dialectic of Credibility and Correctness The Paradox of Evidence and Informativeness Argumentum ad Verecundiam Foxes and Hedgehogs The Asymmetry of Trust and Distrust Cognitive Biases
Substantiation Authority Source Dynamism Trustworthiness Record CREDIBILITY S.T.A.R.S. Substantiation Justification, evidence, explanation Authority Experience, credentials, knowledgeability, mastery Source Dynamism Confident performance, business athlete Trustworthiness Integrity, ethics, reliability, goodwill Record Performance, accuracy, correctness, track record 5
L.I.F.E. LAUNCH INTELLIGENT FORECAST ENGINE EMBRACING THE SCIENCE AND THE ART OF PREDICTION
Intelligence Action Data LAUNCH INTELLIGENT FORECAST ENGINE I have years of historical data, but no easy way to integrate it into my forecast Action I want to complete my forecasting cycle with an accurate baseline number and track the assumptions that were made Intelligence I want to learn from my data and become an empowered forecaster Distinct historic data patterns ingested Hidden insights uncovered Models tested against real data Outputs modified by your experience Assumptions tracked Scenario Planning Growth & revenue potential Forecasting with the best of human and machine Trends quickly surfaced Robust evaluation The Launch Intelligent Forecast Engine improves business forecasts through the integration of machine modelling with human judgment into a structured process of prediction, assessment, and feedback. 6
L.I.F.E. SUBSTANTIATION Analyze historical trend and seasonal patterns, test different forecast models, and apply intelligent model selection 7
L.I.F.E. TRUSTWORTHINESS Launch’s people driven commitment to integrity, transparency, and goodwill
Jared “Professor Madness” Endicott L.I.F.E. AUTHORITY Jared “Professor Madness” Endicott Principal Data Scientist Advanced Analytics Manager AI/ML Solution Architect Launch Consulting 12+ years building predictive models Certified Professional in Demand Forecasting Member of the International Institute of Forecasters Published case study in Hans Levenbach’s “Change & Chance Embraced” Change & Chance Embraced CASE STUDY: A PEER Demand Forecasting Process for Turkey Dinner Cost (Prepared by Jared Endicott following his participation in a CPDF Workshop in Istanbul.) 9
L.I.F.E. RECORD Blend measures of precision and bias, and benchmark forecasts against alternatives to achieve optimized outcomes 10
Have the confidence to own your performance L.I.F.E. SOURCE DYNAMISM Have the confidence to own your performance
Energy Demand Forecasts L.I.F.E. USE CASES Energy Demand Forecasts Data Source: Energy Information Administration (EIA) Business Problem: Forecasts of energy utilization can vary significantly from month to month, year over year, and across locations and sectors. Inaccurate demand forecasts cause inefficient planning. Solution: Automated monthly predictions of the retail sales of electricity, which capture seasonality and trend for each combination of location and sector. Business Value: Increased the precision of 78% of 284 distinct forecasts, for an average improvement of 15%. Increased the credibility and thus the utility of forecasts. Better planning drove greater energy efficiency. 12
Potential Energy Forecast electricity production and load DATA SCIENCE USE CASES Potential Energy Forecast electricity production and load Forecast financials, sales and revenue Root-cause analysis Social listening and micro-targeting NLP and sentiment analysis Customer segmentation Scenarios and simulations Data integration and cleaning 13
THANK YOU
Jared Endicott Principal Data Scientist AI/ML Solution Architect Advanced Analytics Manager JEndicott@LaunchCG.com 206-949-0134 15