Introducing Oracle Crystal Ball Predictor: a new approach to forecasting in MS Excel Environment



Similar documents
CB Predictor 1.6. User Manual

Risk Analysis Overview

<Insert Picture Here> Working With Dashboards for Risk Measurement

IBM SPSS Forecasting 22

Call Centre Helper - Forecasting Excel Template

IBM SPSS Forecasting 21

One-Minute Spotlight. The Crystal Ball Forecast Chart

Spreadsheet software for linear regression analysis

Better decision making under uncertain conditions using Monte Carlo Simulation

Introduction to Statistical Computing in Microsoft Excel By Hector D. Flores; and Dr. J.A. Dobelman

Market Size Forecasting Using the Monte Carlo Method. Multivariate Solutions

Building and Using Spreadsheet Decision Models

Model-driven Business Intelligence Building Multi-dimensional Business and Financial Models from Raw Data

Integrated Resource Plan

Figure 1. An embedded chart on a worksheet.

ModelRisk for Insurance and Finance. Quick Start Guide

Neural Network Add-in

Promotional Forecast Demonstration

DIRECT MAIL: MEASURING VOLATILITY WITH CRYSTAL BALL

Using simulation to calculate the NPV of a project

Activity 3.7 Statistical Analysis with Excel

MODULE 7: FINANCIAL REPORTING AND ANALYSIS

Better planning and forecasting with IBM Predictive Analytics

forecast modeling Statistical Calculations, Replacement Items, and Forecast Templates for Demand Planning Industries Required Modules:

Simulation and Lean Six Sigma

Data Analysis Tools. Tools for Summarizing Data

Using Excel for Handling, Graphing, and Analyzing Scientific Data:

ORACLE HYPERION PLANNING

Excel Companion. (Profit Embedded PHD) User's Guide

SMB Intelligence. Budget Planning

Optimization: Continuous Portfolio Allocation

Extracts from Crystal Ball Getting Started Guide

Prerequisites. Course Outline

BusinessObjects Planning Excel Analyst User Guide

MICROSOFT EXCEL FORECASTING AND DATA ANALYSIS

Excel Add-ins Quick Start Guide

Simple Predictive Analytics Curtis Seare

Downloading RIT Account Analysis Reports into Excel

Polynomial Neural Network Discovery Client User Guide

SAS Add-In 2.1 for Microsoft Office: Getting Started with Data Analysis

Microsoft Access Rollup Procedure for Microsoft Office Click on Blank Database and name it something appropriate.

ADD-INS: ENHANCING EXCEL

sensitivity analysis. Using Excel 2.1 MANUAL WHAT-IF ANALYSIS 2.2 THRESHOLD VALUES

Simulation and Risk Analysis

Financial Planning and Analysis Using Excel

business statistics using Excel OXFORD UNIVERSITY PRESS Glyn Davis & Branko Pecar

Implementing a Customer Lifetime Value Predictive Model: Use Case

Graphing Parabolas With Microsoft Excel

Using Excel for Data Manipulation and Statistical Analysis: How-to s and Cautions

E x c e l : Data Analysis Tools Student Manual

Inquisite Reporting Plug-In for Microsoft Office. Version 7.5. Getting Started

Ansur Test Executive. Users Manual

COMP6053 lecture: Time series analysis, autocorrelation.

Credit Risk Stress Testing

Spreadsheets and Laboratory Data Analysis: Excel 2003 Version (Excel 2007 is only slightly different)

Microsoft Excel. Qi Wei

ORACLE PLANNING AND BUDGETING CLOUD SERVICE

How To Create A Report In Excel

Oracle Planning and Budgeting Cloud Service

Monte Carlo analysis used for Contingency estimating.

An Introduction to Point Pattern Analysis using CrimeStat

Integrating Financial Statement Modeling and Sales Forecasting

EXCEL Tutorial: How to use EXCEL for Graphs and Calculations.

Self-Service Business Intelligence

uncommon thinking ORACLE BUSINESS INTELLIGENCE ENTERPRISE EDITION ONSITE TRAINING OUTLINES

EcOS (Economic Outlook Suite)

Data representation and analysis in Excel

Analytics with Excel and ARQUERY for Oracle OLAP

Excel Guide for Finite Mathematics and Applied Calculus

SQL Server Business Intelligence

Business Portal for Microsoft Dynamics GP Key Performance Indicators

SIMPLIFIED PERFORMANCE MODEL FOR HYBRID WIND DIESEL SYSTEMS. J. F. MANWELL, J. G. McGOWAN and U. ABDULWAHID

seven Statistical Analysis with Excel chapter OVERVIEW CHAPTER

Oracle Hyperion Planning

TheFinancialEdge. Reports Guide for General Ledger

Session 10. Laboratory Works

ISSUES IN UNIVARIATE FORECASTING

ELECTRO-MECHANICAL PROJECT MANAGEMENT

Data Mining for Business Intelligence. Concepts, Techniques, and Applications in Microsoft Office Excel with XLMiner. 2nd Edition

How to use MS Excel to regenerate a report from the Report Editor

YASAIw.xla A modified version of an open source add in for Excel to provide additional functions for Monte Carlo simulation.

DPL. Portfolio Manual. Syncopation Software, Inc.

REAL OPTIONS VALUATION, INC.

Causal Leading Indicators Detection for Demand Forecasting

STATISTICAL ANALYSIS WITH EXCEL COURSE OUTLINE

The power of IBM SPSS Statistics and R together

Anytime 500 Forecast Modeling

Forecasting in STATA: Tools and Tricks

Below is a very brief tutorial on the basic capabilities of Excel. Refer to the Excel help files for more information.

Interrupted time series (ITS) analyses

Forecasting Stock Prices using a Weightless Neural Network. Nontokozo Mpofu

Using Excel for Statistical Analysis

Preface of Excel Guide

Evaluating Trading Systems By John Ehlers and Ric Way

PortfolioCenter Export Wizard in Practice: Evaluating IRA Account Holder Ages and Calculating Required Minimum Distribution (RMD) Amounts

MS Project Tutorial for Senior Design Using Microsoft Project to manage projects

Transcription:

<Insert Picture Here> Introducing Oracle Crystal Ball Predictor: a new approach to forecasting in MS Excel Environment Samik Raychaudhuri, Ph. D. Principal Member of Technical Staff ISF 2010

Oracle Crystal Ball An add-in to Microsoft Excel for performing: Monte Carlo simulation Stochastic optimization Time series forecasting The focus of the software has been historically on Monte Carlo simulation, with basic time-series forecasting capabilities The current version finally provides much-needed attention to the forecasting capabilities by introducing an array of features to the tool

Oracle Crystal Ball Predictor Features The time-series forecasting tool in Oracle Crystal Ball is called Predictor. We will refer to it henceforth as CB Predictor The tool has an interesting set of usability features which sets it apart from other forecasting software: Works completely in the familiar Microsoft Excel spreadsheet environment: no data import or result export required Officially supported on MS Excel XP, 2003 and 2007. Excel 2010 will be supported soon Usual analytical features available (details follow) Ease of use, with non-intimidating dialogs and sensible defaults for the uninitiated and providing array of configuration options for power users Professional forecasting charts and reports Seamlessly integrates with Monte Carlo simulation for conducting risk analysis along with time-series forecasting Can definitely be the forecasting software of choice for the rest of us!!

CB Predictor Analytical Feature List CB Predictor has an extensive list of features designed to make the forecasting experience easy and productive The tool sports a wizard like interface to guide users through the forecasting process The features can be broadly subdivided into two categories: Data preparation and forecasting Result analysis and reporting

CB Predictor Analytical Feature List: Data Preparation and Forecasting Data preparation and forecasting is performed over four feature-rich screens Identifying input data Describing data characteristics Selecting models to run Choosing forecasting options

Data Preparation and Forecasting: The Welcome Screen CB Predictor starts off at the Welcome screen Talks about the basic forecasting procedure

Data Preparation and Forecasting: Identifying Input Data

Data Preparation and Forecasting: Identifying Input Data (Contd.) The hidden features Intelligent data selection Select one cell in a contiguous range of series and start CB Predictor Identifies the complete range of data, orientation of the data, and the position of header and date ranges if exist Supports discontinuous data range (e.g., alternate rows or columns of data) Supports logical aligning of data: pre-data gaps Automatically identify various type of periods like months in an year, dates, or quarters etc.

Data Preparation and Forecasting: Describing Data Characteristics [Planned Screen] Seasonality detection Events modeling (New) Missing value imputation Outlier detection

Data Preparation and Forecasting: Describing Data Characteristics (Contd.) Seasonality detection We automatically detect seasonality for input series One can override the seasonality of each series or set them to non-seasonal The detection algorithm uses threshold-based analysis of autocorrelation and their probabilities at various lags in the data Has been tested extensively on M1-competition and M3-competition data

Data Preparation and Forecasting: Describing Data Characteristics (Contd.) Data screening Missing value imputation: CB Predictor can impute the missing values in the dataset Uses nearest neighbor interpolation or cubic spline interpolation Options to control the interpolation scheme Outlier detection: detects outliers specific to each forecasting method Suggests replacing values Options to control the method and the aggressiveness of the detection algorithm In each case, charts are available to ease the decision making process Defaults work great for majority of scenarios

Data Preparation and Forecasting: Describing Data Characteristics (Contd.) Events modeling Still in the works, slated for a future release

Data Preparation and Forecasting: Selecting Forecasting Models [Planned Screen]

Data Preparation and Forecasting: Selecting Forecasting Models (Contd.) Non-seasonal Models Single moving average Single exponential smoothing Double moving average Double exponential smoothing Seasonal Models Seasonal Additive Seasonal Multiplicative Holt-Winters Seasonal Additive Holt-Winters Seasonal Multiplicative Order and other parameters are automatically detected or can be overridden by the user

Data Preparation and Forecasting: Selecting Forecasting Models (Contd.) Multiple Linear Regression Supports lagged dependent variables Supports multiple dependent variables Stepwise regression (forward and iterative) for choosing important independent variables from a pool Performs automatic forecasting of the dependent variable using regression equation and forecasts from independent variable ARIMA (New) Still in the works, slated for a future release

Data Preparation and Forecasting: Forecasting Options

Features: CB Predictor Analytical Feature List: Result Analysis and Reporting Displays forecast and confidence intervals Displays the best method for each series with easy browsing for other series and methods Important statistics for each method Seasonal bands for visual identification of patterns in the historical and forecasting horizon Adjust forecasting horizon and CI on-the-fly

CB Predictor Analytical Feature List: Result Analysis and Reporting (Contd.) Manual adjustment of forecasts Supports various type of adjustments

CB Predictor Analytical Feature List: Result Analysis and Reporting (Contd.) Access to reports and data extraction from the results window menu Integration with CB Monte Carlo simulation Forecasts are treated as normal probability distributions with forecast value as the mean and standard error as the standard deviation Can then be used for risk analysis

Example: Using CB Predictor for Forecasting and Risk Analysis: Monica s Bakery A rapidly-growing boutique bakery in Taos, New Mexico The owner, Monica, has kept records of sales of her three main products: French bread, Italian bread and pizza Wants to analyze the cash flow of the business to purchase fixed assets

Example: Monica s Bakery (Contd.) Demonstration Reports

Example: Monica s Bakery (Contd.) Static analysis shows: Doesn t look good as per the minimum cash target goes But is it really the true picture? Risk analysis thinking in ranges

Example: Monica s Bakery (Contd.) The risk analysis of the cash flow Lot of uncertain variables in the spreadsheet model The numbers really represent the likely scenario or the best guess scenario Let s try to analyze the uncertainty in some of these variables and see the interaction effect Set probability distributions on top of input variables (called assumptions) COGS Overhead Financing Taxes Set the target variables in which we are interested (called forecasts) Run a Monte Carlo Simulation 5000 trials Get the probability distribution for the target variables

Example: Monica s Bakery (Contd.) Use the forecast charts to find out the probability of hitting the minimum cash targets Probability for July: 65.37% Probability for August: 30.97% Probability for September: 97.17%

Example: Monica s Bakery (Contd.) How about considering the uncertainty in the forecast values themselves? Have those as assumptions as well New probability of hitting the minimum cash targets Probability for July: 64.42% Probability for August: 33.24% Probability for September: 96.55% These are when you have chosen to use the best forecasting method, which has a reasonably narrow CI Choose another forecasting methods having wider forecast CI s and verify that the risk spread increases for target variables

Questions? Presenter: Samik Raychaudhuri samik.raychaudhuri@oracle.com