Anurag Seksaria, profile picture
Uploaded byAnurag Seksaria
92 views

Overview business analytics 13th june 2020 v4

This document outlines an agenda for a business analytics overview presentation. The session plan includes an introduction to customer segmentation using a case study, exploring and modeling data through the case study, interpreting results and managing change, opportunities in AI and ML, unstructured analytics, comparing software tools, and expanding analytics to multiple areas through case studies. The document also provides background information on BMGI, an analytics company, including its mission, services, industries served, locations, employees, and publications. Key aspects of business analytics such as the four analysis types and examples are defined. Finally, a case study on predicting personal loan acceptance is described, including the data structure.

Embed presentation

1 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Business Analytics Overview 13th June 2020 Anurag Seksaria anurags@BMGIndia.com
2 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Session Plan 1. About BMGI 2. Why Business Analytics & Need for Structured Approach 3. Introduction to Customer Segmentation Case 4. SCRUB, EXLPLORE & MODEL Data – Through Case Study 5. INTERPRET & Manage Change 6. AI & ML Opportunities 7. Unstructured Analytics 8. Software Tools Comparison 9. Expanding to Multiple Areas & Case Studies
3 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. OUR MISSION “To partner with our clients for unlocking their inherent potential through innovative solutions and delivering breakthrough results”
4 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. BMGI OVERVIEW INSOURCING LEADERSHIP OBJECTIVITY INSIGHT GLOBAL EMPLOYEES100+ GEOGRAPHIES SERVICED16+ CLIENTS SERVICED200+RESULTS BMGI enables businesses to solve strategic, organizational and process problems. COMPETENCIES § Innovation and Design § Customer Insighting § Strategy Planning & Deployment § Problem Solving § Process Improvement INDUSTRIES • Services. • IT / ITeS / BPO • Education • BFSI • Hospitality PUBLICATIONS § The Innovator’s Toolkit § Insourcing Innovation § The Complete Idiot’s Guide to Lean Six Sigma § Design for Lean Six Sigma Manufacturing • Pharma • FMCG • Chemicals • Automotive
5 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. BMGI’S Services For Enabling Business Success StrategizeInnovate Solve Problems Transform Business
6 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. WHAT CAUSED THE CHANGE Significant Reduction in Accidents due to drunken driving 30% Move from Self Driving to UBER Local Police has become Tough on Drunken Drivers Drinking itself has reducing
7 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Why UPS Trucks Never Take Left Turn
8 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. § Business analytics (BA) refers to the skills, tools & techniques, practices for continuous iterative exploration and investigation of past business performance to gain insight and drive improvements § Business analytics is based on data and statistical methods. What is Business Analytics?
9 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Machine Learning – Training – Testing - Validation Original Set Training Set Training Set Testing Set Testing Set Validation. Set Machine Learning Algorithmic Predative Model
10 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Four Pillars of Business Analytics Studies Ø Descriptive Analytics - What Happened? Ø Diagnostic Analytics - Why Did it Happen? Ø Predictive Analytics - What Happens If? Ø Prescriptive Analytics - How to Make it Happen?
11 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. üHow much volume did we sell last month in the GREEN ZONES? üHow many of our consignments reached/arrived late? üHow much was the delay in receiving the payments? üHow much is the error rate or rejection rate? Descriptive Analytics - What Happened?
12 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. üA sudden drop/increase in sales? üSudden increase in customer returns or Internal Rejection Rates? üHigher number of employee taking leave? Look for cause and effect to illustrate why something occurred Techniques include classification and regression analysis (CART) Diagnostic Analytics - Why Did it Happen?
13 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Beer and Diapers
14 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Beer and the Diapers v Wal-Mart decided to combine the data from its loyalty card system (demographic Information) with its point of sale systems (what customer bought) v Once combined, the data was mined extensively and many correlations appeared. v People who buy GIN in are also likely to buy TONIC, they often also buy LEMONS v However, one correlation stood out v On Friday afternoons, young American who buy DIAPERS (nappies) also buy BEER
15 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Strategic and Incidental Purchase Ø Bread & Butter/Cheese Ø Beer and Chips How should we place them
16 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. üIf the monsoon is delayed by a month, grains process might go up by xx% üIf the competitor increases the discount by x% our sales will go down by y% üIf RBI reduces interest rates by xx%, stock market will…. üIf we reduce our marketing spend by Rs xyz, we will loose sales by xx% ê Combine outcomes of descriptive and diagnostic analytics into actionable insights for decision making ê Techniques include - Confidence intervals, T statistics and P values, applied machine learning algorithms, classification models and regression models Predictive Analytics - What Happens If
17 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. üIf I use higher grade ingredient the yield will improve by x% üIf We increase the bulk discount by x%, sales will go up by y% üIf we increase the machine speed by 25%, rejections are likely to go up by 1% ê Techniques include - Artificial intelligence, Machine Learning and Neural Network algorithms Prescriptive Analytics - How to Make it Happen
18 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Potential Value Delivery through Analytics? § Optimization • What price should we charge for this product? • How much inventory should we keep? § Segmentation • Who should we market to? • More importantly – who should we NOT market to? § Recommendations • You bought X therefore you may want to buy Y? • Right set of employees for given type of Roles? § Forecasting • How much will we sell next quarter? • Repeating weather patterns? • Volumes by Day of The Week, Day of the Month? • Attrition by Level of Employees § Risk Assessment • Who will pay me, on time? Who will not? • Potential Fraud
19 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Structured vs Unstructured Data Ø Structured § Tables with rows and columns § Each data set follows a schema/structure Ø Unstructured § Videos § Images § Documents § No schema/structure, but data is there
20 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Why is Analytics Gaining Importance A. Availability of Good Data B. Computing Capability C. Higher number of influencing Factors D. Dynamic Environment and Need to Decide Fast E. Need to establish insights from unknown & unexplored factors
21 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. OSEMN Methodology Overview Public/Web data Cloud Database Centralized database Local machines Obtain Scrub Explore Model iNterpret Cleaning data • Joining data • Standardizing • Missing Values & Duplicates • Data Formatting • Feature Scaling • Data Encoding Exploratory Analysis • Descriptive Statistics • Graphical Analysis • Comparing Groups • Feature Engineering Regression Decision Tree Logistic Regression Cluster Analysis Naive Bayes classifier Data Interpretation Hyperparameter Tuning Iterations Delivering Business Value Soft Skills • Stakeholder Analysis • Communication Plan • Questioning Techniques • Change Management Case Studies
23 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. BIG Data needs special attention The term ‘Big Data’ is: Ø Frequently abused….. Ø Thrown around by bloggers and journalists to increase hits…. Ø Data ‘too big to fit in Excel’ – Really???
24 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. What is Big Data: 5 Vs Variety – different types of data Velocity – speed at which new data is generated Volume – amount of data Veracity – lack of quality and accuracy VALUE – turn data into information, into knowledge, into wisdom
25 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Where does all this Big Data come from? What's important for me, for my organization, my customer? • Immediate • Short Term • Mid Term • Long Term
26 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Supervised and Unsupervised Learning Unsupervised Supervised
27 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Supervised Learning Characteristics of Employees Who Buy Our Product Yes - Buy No - Do Not Buy
28 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Types of Learning - Supervised • Training data includes desired outputs. • Machine is trained on past data and its output to learn the hidden pattern. Hence Supervised
30 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Unsupervised Learning Segment Employee with known Characteristics Oval Body Oval Body & Square Head Oval Body & Circular Head Oval & Gray Body with Circular Head Oval & White Body with Circular Head
31 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Types of Learning - Unsupervised • Training data does not include desired outputs • Machine is not trained on any specific output. Hence, the name Unsupervised Learning
33 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Types of Data Ø Variable or Continuous § Salary in Dollars, Number of Years of experience Ø Discreate or Attribute § Binary - Yes or No, Good or Bad, Present or Not Present § Ordinal - No of Family Members § Categorical – ZIP Code, Education, Profession
34 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Machine Learning Unsupervised Learning Reinforcement LearningSupervised Learning Target Not Available Target Continuous Data Target Categorical Data Target Categorical Data Target Not Available Regression Classification Clustering Association Classification Control Use Cases: • Ad Popularity • Predicting Prices • Growth Prediction Tabular Data: Audio / Video Data: Use Cases: • Predicting Churn • Fraud Detection • User Conversion Tabular Data: Audio / Video Data: Use Cases: • Segmentation • Recommender Systems Tabular Data: Audio / Video Data: Use Cases: • Market basket analysis Tabular Data: Audio / Video Data: Use Cases: • Real-time Decisions • Gameplay • Learning Skills Tabular Data: Audio / Video Data: Use Cases: • Self Driving Cars • Fault Detection • Robot Navigation Tabular Data: Audio / Video Data:
35 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Background of the Project Probability of taking Personal Loan q A retail bank is interested in finding out the influence of Cross Selling Efforts to the existing customers w.r.t. decision to take a Personal Loan q The data were collected on various influencing factors given in next slide
36 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Data Structure 1 of 2 ü Personal Loan - Did this customer accept the personal loan offered in the last campaign? ü Customer ID ü Customer's age in completed years ü Experience #years of professional experience ü Income Annual income of the customer ($000) ü ZIPCode Home Address ZIP code. ü Family size of the customer ü Credit Card Average spending on credit cards per month ü Education Level – Non-Graduate, Graduate; Professional
37 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Data Structure 2 of 2 ü Mortgage Value of house mortgage if any. ($000) ü Securities Account Does the customer have a securities account with the bank? ü Certificate of Deposit (CD) account with the bank? ü Online Does the customer use internet banking facilities? ü Credit Card – Yes or No ü Does the customer use a credit card issued by UniversalBank?
38 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Big Data with Large Number of Features is Often Unbalanced Number of Customers by ZIP Code & Education Education ZIP Codes 1 2 3 Grand Total 9307 1 1 90007 5 1 6 90009 4 2 2 8 90011 1 1 1 3 90016 2 2 90018 2 2 4 90019 1 2 2 5 90024 20 12 18 50 90025 5 7 7 19 90027 2 1 3 90028 4 2 2 8 90029 4 1 5 90032 5 2 1 8 90033 4 2 3 9 90034 12 2 6 20 90035 1 1 4 6 90036 5 1 1 7 90037 2 1 2 5 90041 5 1 4 10 90044 1 1 2 90045 3 3 Number of Customers 1. Missing Not at Random (MNAR) 2. Missing at Random (MAR) 3. Missing completely at Random (MCAR) 4. Missing at Random (MAR) Missing completely at Random (MCAR)
39 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works.
40 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Ø Missing Values Ø Duplicate Value Ø Dimensionality/Scale Ø Outliers/Extreme Values Ø Combining Data from Multiple Sources SCRUB
41 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. EXPLORE : How reliable is Analysis on Pivot Table…. No of Family Member % Loan Availed No Yes 1 93% 7% 2 92% 8% 3 87% 13% 4 89% 11% Grand Total 90% 10% Education % Loan Availed No Yes Non Graduate 96% 4% Graduate 87% 13% Post Graduate 86% 14% Grand Total 90% 10% % Avail Loan Family Size Education 1 2 3 4 Overall Non Graduate 1% 1% 11% 10% 4% Graduate 12% 19% 11% 11% 13% Post Graduate 12% 14% 18% 12% 14% Overall 7% 8% 13% 11% 10%
42 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. EXPLORE - Data Distributions
43 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Complex relationships between Output and Inputs Number of relationships cannot be fitted with Liner, Quadratic or even advance polynomial equations
44 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. MODEL - CART
45 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Cost of Misclassification
46 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Relative Importance of Factors
47 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Decision Tree Overall Hit Rate 9.5% Salary < 92.5K 0.7% Income>92.5% 33% Non Graduates 10% Graduates + 72% Family 1,2 Near 0% Family 3+ 38% Focus on Clients with • Income >92.5K • Graduates + • Non-Graduates - Family Size 3+
48 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Train and Test Model Summary
49 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Prediction q The model provides ability to predict hit rate for multiple profiles q Individuals with income 150K & Non-Graduates, the hit rate is below 1% if family size is 1 or 2, this goes up to 72% if family size is 3 or 4.
50 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Confusion – Train v/s Test
51 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Unsupervised Learning – Cluster Observations q Scenarios where no specific outcome is specified or known q In the same banking environment we have information of 5000 customers profiles. q Objective is to Segment customers in the logical groups
52 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Cluster Observations - Dendrogram
53 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Cluster Centroids Variable Cluster1 Cluster2 Cluster3 Cluster4 Cluster5 Grand centroid Age 43.4 46.1 46.7 42.1 45.7 45.3 Experience 18.1 21.0 21.4 17.0 20.7 20.1 Income 29.3 108.2 66.1 188.6 149.6 73.8 Family 2.5 2.1 2.6 2.0 2.0 2.4 CCAvg 0.9 2.6 1.7 4.7 4.0 1.9 Education 2.0 1.6 2.0 1.7 1.5 1.9 Cluster Properties – Income and Credit Card Spend No of Observations. 1998 499 1761 104 158 5000 Chances of Taking Loan. 0%. 29%. 3%. 51%. 43%
54 Unstructured Analytics Sourcing Extracting Categorizing Visualization
55 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Customer Sentiment Analysis
56 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Tracing LinkedIn Profile Application ü Swipe in swipe out ü New Employee Onboarding ü Guest/Visitor Management
57 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Security Enhancement
58 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. SPACE CUSTOMER FEEDBACK – RESORT BEACH ROOM Far Crowd Water Room Service Fun Great Shower Staff Clean
59 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. IShallNever MixWater Prof Juran’s Example on Root Cause Analysis
60 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Threat to Causal Inferences 1. Selection 2. Simultaneity 3. Omitted Variable 4. Measurement Error
61 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Selection – Sample is Not Representative
62 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Correlation Causation
63 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Simultaneity – X and Y Cause Each Other
64 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Point to Remember q Any such analysis should be used with complete care as misinterpretations can add to frustration and result in disbelief in scientific methods q While we are using advanced analytics tools to further strengthen our ability to identify root cause, Team Engagement, Brainstorming, FMEA, Why-Why Analysis, 6-3-5, Mistake Proofing, Daily Work Management remain equally important q Good old Rule applies Business Problem – Statistical Problem – Statistical Solution – Business Solution
65 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Data Enrichment - Using Secondary Data Data enrichment is the process of creating new features by introducing data from external sources. ü Information on Bank’s Customer Salary Increase Decrease can be extracted from Saving Bank Transaction Records ü Expenditure pattern can be extracted from Credit Card Bills ü Addition in family member can be accessed through Birth and Death Register, Mediclaim’s, etc.
66 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Opportunity to Use AI and ML q The model can be linked to multiple data sources within/outside bank to update the information on existing parameters and add additional parameters like Age of Family Members, Changes in Income and investment, etc q This information can be used to auto trigger the agents to enable enhancement of cross selling effectiveness, launching new products etc.
Licensing Open Source Open Source SaaS High / One-Time SaaS Ease of use Difficult (Requires Coding) Difficult (Requires Coding) Difficult (Requires Coding) Medium Easy Applications Broad Broad Broad Narrow Dataset Specific Easy to Understand Difficult (Requires Coding) Difficult (Requires Coding) Difficult (Requires Coding) Medium Easy Setup Time Months Months Weeks Weeks 10 mins Training Speed Weeks / Months Weeks / Months Days Days < 1 minute Deployment Time Months Months Weeks Weeks < 1 minute Engineer Involvement High (Requires Coding) High (Requires Coding) High (Requires Coding) Medium None Cloud Specific? No No Yes (Only Supports AWS) No No Support N/A N/A via Email via Email 24x5 Hands-On Support Software Landscape & Comparison
68 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. BMGI Certified Business Analytics Professional Module 1 : Foundation Module 2 : Advanced Modelling + Python & R (Optional) Module 3 : Forecasting & Visualisation Module 4 - Optimisation, AI & ML
69 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Case Studies
70 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Sample Case Studies 1. Reduction in Processing Time and Error Rate 2. Profitability improvement in multi channel Sales 3. Predicting and reducing days of Sales Outstanding 4. Deriving actionable insights from Employee Satisfaction Survey 5. Reduction in Customer Returns and Complaints 6. Reduction in Procure to Pay Cycle Time
71 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works.
72 © BMGI. Except as may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Wish you All the Best!!!!

More Related Content

PDF
Overview business analytics 6th june 2020 v5
byAnurag Seksaria
 
PPTX
India's Leading Data Analytics Partner: Empowering Businesses with BMGI
byNaresh Raisinghani
 
PDF
BMGI India's Leading Data Analytics Partner Empowering Businesses with BMGI.pdf
byNaresh Raisinghani
 
PDF
Maximizing Growth and Efficiency_ The Impact of Advanced Analytics & Continuo...
byNaresh Raisinghani
 
PPTX
Continuous Improvement Through Data Insights_ How BMGI India’s Advanced Analy...
byNaresh Raisinghani
 
PDF
Continuous Improvement Through Data Insights_ How BMGI India’s Advanced Analy...
byNaresh Raisinghani
 
PPTX
Maximizing Growth and Efficiency: The Impact of Advanced Analytics & Continuo...
byNaresh Raisinghani
 
PPTX
Business Analytics, Types of Business Analytics
byRoshanBhattarai17
 
Overview business analytics 6th june 2020 v5
byAnurag Seksaria
 
India's Leading Data Analytics Partner: Empowering Businesses with BMGI
byNaresh Raisinghani
 
BMGI India's Leading Data Analytics Partner Empowering Businesses with BMGI.pdf
byNaresh Raisinghani
 
Maximizing Growth and Efficiency_ The Impact of Advanced Analytics & Continuo...
byNaresh Raisinghani
 
Continuous Improvement Through Data Insights_ How BMGI India’s Advanced Analy...
byNaresh Raisinghani
 
Continuous Improvement Through Data Insights_ How BMGI India’s Advanced Analy...
byNaresh Raisinghani
 
Maximizing Growth and Efficiency: The Impact of Advanced Analytics & Continuo...
byNaresh Raisinghani
 
Business Analytics, Types of Business Analytics
byRoshanBhattarai17
 

Similar to Overview business analytics 13th june 2020 v4

PDF
BizTrans SysTech_Analytics_Serv_SAP_v1.0
byBizTrans SysTech
 
PPTX
Introduction to Business Analytics MBU880.pptx
byChintalapatiNeelimaR
 
PPTX
Introduction to Business Data Analytics
byVadivelM9
 
PPTX
Unit 1.pptx Anna University Business Analytics
byssuserff6fd4
 
PPTX
Chapter 1 - Introduction to Business Analytics.pptx
byAhmadM65
 
PPTX
1_2 Business Analytics Overview.pptx
byrobert141042
 
PPTX
Ba syllabus
byBiraSurannavar1
 
DOC
BIG DATA & BUSINESS ANALYTICS
byVikram Joshi
 
PPT
Introduction to Business Analytics and Tools
byKristu Jayanti College, Bengaluru
 
PDF
LESSON 1.pdf
bycalf_ville86
 
PDF
Using data analytics to drive BI A case study
byPhoenixraj
 
PDF
BMGI India Addressing Strategic, Innovative, and Operational Problems at Core...
byNaresh Raisinghani
 
PPTX
BMGI India: Transforming Business Challenges into Strategic Advantages
byNaresh Raisinghani
 
PDF
Big data-analytics-ebook
byShubhashish Biswas
 
PPTX
1.1. Evolution-and-Scope-of-Business-Analytics.pptx
byJitendra Tomar
 
PDF
BMGI India: Transforming Business Challenges into Strategic Advantages
byNaresh Raisinghani
 
PPTX
Mastering Business Analytics - navigating skills and opportunities
byUtsab18
 
PDF
Ibm
byebuc
 
PDF
Business Analytics
byMeganShaw38
 
PDF
Introduction to Business Analytics.pdf
bySuku Thomas Samuel
 
BizTrans SysTech_Analytics_Serv_SAP_v1.0
byBizTrans SysTech
 
Introduction to Business Analytics MBU880.pptx
byChintalapatiNeelimaR
 
Introduction to Business Data Analytics
byVadivelM9
 
Unit 1.pptx Anna University Business Analytics
byssuserff6fd4
 
Chapter 1 - Introduction to Business Analytics.pptx
byAhmadM65
 
1_2 Business Analytics Overview.pptx
byrobert141042
 
Ba syllabus
byBiraSurannavar1
 
BIG DATA & BUSINESS ANALYTICS
byVikram Joshi
 
Introduction to Business Analytics and Tools
byKristu Jayanti College, Bengaluru
 
LESSON 1.pdf
bycalf_ville86
 
Using data analytics to drive BI A case study
byPhoenixraj
 
BMGI India Addressing Strategic, Innovative, and Operational Problems at Core...
byNaresh Raisinghani
 
BMGI India: Transforming Business Challenges into Strategic Advantages
byNaresh Raisinghani
 
Big data-analytics-ebook
byShubhashish Biswas
 
1.1. Evolution-and-Scope-of-Business-Analytics.pptx
byJitendra Tomar
 
BMGI India: Transforming Business Challenges into Strategic Advantages
byNaresh Raisinghani
 
Mastering Business Analytics - navigating skills and opportunities
byUtsab18
 
Ibm
byebuc
 
Business Analytics
byMeganShaw38
 
Introduction to Business Analytics.pdf
bySuku Thomas Samuel
 

Recently uploaded

PDF
Azure Data Factory Overview for SSIS Developers.pdf
bySrinivas V
 
PPTX
Statistical Analysis for Researchers - Presentation - Lecture Two.pptx
byAmgad Fahmy
 
PDF
AI 21.pdf Learn How AI Tools Optimize Decision Making, Marketing, HR, and More
byCA Suvidha Chaplot
 
PPTX
Sales_Analysis_Presentation.pptx dashboard
bydheerendras984
 
PDF
AI 1.pdf AI in Business Training - Revolutionizing Business Operations
byCA Suvidha Chaplot
 
PDF
Douglas-Peucker Algorithm : how works in Deswik
by😀 Shahrokh Paravarzar, PhD, E.I.T
 
PPTX
Investment and Portfilio. Group no 1.pptx
byAhtshamUlHaq10
 
PDF
AI 30.pdf Practical Insights from CA Suvidha Chaplot for AI Implementation in...
byCA Suvidha Chaplot
 
PPT
unit-3 Introduction to drawing and instruments part-2.ppt
byalemayehuc1
 
PPTX
Ch6 Automata & Complexity Theory 2016.pptx
bytechnologyn387
 
PPTX
25CSM2R26_AditiMIshra_WDSAttackcybe.pptx
bydrfareesafirdouse
 
PPTX
Ollama_RAG_retrivewantedvaluesusingModels.pptx
byanjaneyav00
 
PPTX
Testing-Results-and-Outcomes (1).pptx dfw
bySrini Vasan
 
PDF
CSE550_ Lecture 02robosssssssssssstik.pdf
byfarfauziar
 
PPTX
RAG_classnoteswith having various ways of publishing the model.pptx
byanjaneyav00
 
PPTX
GLOBAL TREND CH-2 Power Point new chapter .pptx
bytechnologyn387
 
PDF
Airport Familiarization.pdffffffffffffffff
byamreentahamaqbool8b
 
PPTX
2ai_mockinterview.pptx hjijjjjjjjjjnhhbh
bypankajponia112
 
PPTX
LinkedIn Lead Generation expert for B2B Lead Generation
byMdBelalSharif1
 
PDF
data data data data data data data data data
byMarwan Hadid
 
Azure Data Factory Overview for SSIS Developers.pdf
bySrinivas V
 
Statistical Analysis for Researchers - Presentation - Lecture Two.pptx
byAmgad Fahmy
 
AI 21.pdf Learn How AI Tools Optimize Decision Making, Marketing, HR, and More
byCA Suvidha Chaplot
 
Sales_Analysis_Presentation.pptx dashboard
bydheerendras984
 
AI 1.pdf AI in Business Training - Revolutionizing Business Operations
byCA Suvidha Chaplot
 
Douglas-Peucker Algorithm : how works in Deswik
by😀 Shahrokh Paravarzar, PhD, E.I.T
 
Investment and Portfilio. Group no 1.pptx
byAhtshamUlHaq10
 
AI 30.pdf Practical Insights from CA Suvidha Chaplot for AI Implementation in...
byCA Suvidha Chaplot
 
unit-3 Introduction to drawing and instruments part-2.ppt
byalemayehuc1
 
Ch6 Automata & Complexity Theory 2016.pptx
bytechnologyn387
 
25CSM2R26_AditiMIshra_WDSAttackcybe.pptx
bydrfareesafirdouse
 
Ollama_RAG_retrivewantedvaluesusingModels.pptx
byanjaneyav00
 
Testing-Results-and-Outcomes (1).pptx dfw
bySrini Vasan
 
CSE550_ Lecture 02robosssssssssssstik.pdf
byfarfauziar
 
RAG_classnoteswith having various ways of publishing the model.pptx
byanjaneyav00
 
GLOBAL TREND CH-2 Power Point new chapter .pptx
bytechnologyn387
 
Airport Familiarization.pdffffffffffffffff
byamreentahamaqbool8b
 
2ai_mockinterview.pptx hjijjjjjjjjjnhhbh
bypankajponia112
 
LinkedIn Lead Generation expert for B2B Lead Generation
byMdBelalSharif1
 
data data data data data data data data data
byMarwan Hadid
 

Overview business analytics 13th june 2020 v4

  • 1.
    1 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Business Analytics Overview 13th June 2020 Anurag Seksaria anurags@BMGIndia.com
  • 2.
    2 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Session Plan 1. About BMGI 2. Why Business Analytics & Need for Structured Approach 3. Introduction to Customer Segmentation Case 4. SCRUB, EXLPLORE & MODEL Data – Through Case Study 5. INTERPRET & Manage Change 6. AI & ML Opportunities 7. Unstructured Analytics 8. Software Tools Comparison 9. Expanding to Multiple Areas & Case Studies
  • 3.
    3 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. OUR MISSION “To partner with our clients for unlocking their inherent potential through innovative solutions and delivering breakthrough results”
  • 4.
    4 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. BMGI OVERVIEW INSOURCING LEADERSHIP OBJECTIVITY INSIGHT GLOBAL EMPLOYEES100+ GEOGRAPHIES SERVICED16+ CLIENTS SERVICED200+RESULTS BMGI enables businesses to solve strategic, organizational and process problems. COMPETENCIES § Innovation and Design § Customer Insighting § Strategy Planning & Deployment § Problem Solving § Process Improvement INDUSTRIES • Services. • IT / ITeS / BPO • Education • BFSI • Hospitality PUBLICATIONS § The Innovator’s Toolkit § Insourcing Innovation § The Complete Idiot’s Guide to Lean Six Sigma § Design for Lean Six Sigma Manufacturing • Pharma • FMCG • Chemicals • Automotive
  • 5.
    5 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. BMGI’S Services For Enabling Business Success StrategizeInnovate Solve Problems Transform Business
  • 6.
    6 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. WHAT CAUSED THE CHANGE Significant Reduction in Accidents due to drunken driving 30% Move from Self Driving to UBER Local Police has become Tough on Drunken Drivers Drinking itself has reducing
  • 7.
    7 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Why UPS Trucks Never Take Left Turn
  • 8.
    8 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. § Business analytics (BA) refers to the skills, tools & techniques, practices for continuous iterative exploration and investigation of past business performance to gain insight and drive improvements § Business analytics is based on data and statistical methods. What is Business Analytics?
  • 9.
    9 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Machine Learning – Training – Testing - Validation Original Set Training Set Training Set Testing Set Testing Set Validation. Set Machine Learning Algorithmic Predative Model
  • 10.
    10 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Four Pillars of Business Analytics Studies Ø Descriptive Analytics - What Happened? Ø Diagnostic Analytics - Why Did it Happen? Ø Predictive Analytics - What Happens If? Ø Prescriptive Analytics - How to Make it Happen?
  • 11.
    11 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. üHow much volume did we sell last month in the GREEN ZONES? üHow many of our consignments reached/arrived late? üHow much was the delay in receiving the payments? üHow much is the error rate or rejection rate? Descriptive Analytics - What Happened?
  • 12.
    12 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. üA sudden drop/increase in sales? üSudden increase in customer returns or Internal Rejection Rates? üHigher number of employee taking leave? Look for cause and effect to illustrate why something occurred Techniques include classification and regression analysis (CART) Diagnostic Analytics - Why Did it Happen?
  • 13.
    13 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Beer and Diapers
  • 14.
    14 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Beer and the Diapers v Wal-Mart decided to combine the data from its loyalty card system (demographic Information) with its point of sale systems (what customer bought) v Once combined, the data was mined extensively and many correlations appeared. v People who buy GIN in are also likely to buy TONIC, they often also buy LEMONS v However, one correlation stood out v On Friday afternoons, young American who buy DIAPERS (nappies) also buy BEER
  • 15.
    15 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Strategic and Incidental Purchase Ø Bread & Butter/Cheese Ø Beer and Chips How should we place them
  • 16.
    16 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. üIf the monsoon is delayed by a month, grains process might go up by xx% üIf the competitor increases the discount by x% our sales will go down by y% üIf RBI reduces interest rates by xx%, stock market will…. üIf we reduce our marketing spend by Rs xyz, we will loose sales by xx% ê Combine outcomes of descriptive and diagnostic analytics into actionable insights for decision making ê Techniques include - Confidence intervals, T statistics and P values, applied machine learning algorithms, classification models and regression models Predictive Analytics - What Happens If
  • 17.
    17 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. üIf I use higher grade ingredient the yield will improve by x% üIf We increase the bulk discount by x%, sales will go up by y% üIf we increase the machine speed by 25%, rejections are likely to go up by 1% ê Techniques include - Artificial intelligence, Machine Learning and Neural Network algorithms Prescriptive Analytics - How to Make it Happen
  • 18.
    18 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Potential Value Delivery through Analytics? § Optimization • What price should we charge for this product? • How much inventory should we keep? § Segmentation • Who should we market to? • More importantly – who should we NOT market to? § Recommendations • You bought X therefore you may want to buy Y? • Right set of employees for given type of Roles? § Forecasting • How much will we sell next quarter? • Repeating weather patterns? • Volumes by Day of The Week, Day of the Month? • Attrition by Level of Employees § Risk Assessment • Who will pay me, on time? Who will not? • Potential Fraud
  • 19.
    19 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Structured vs Unstructured Data Ø Structured § Tables with rows and columns § Each data set follows a schema/structure Ø Unstructured § Videos § Images § Documents § No schema/structure, but data is there
  • 20.
    20 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Why is Analytics Gaining Importance A. Availability of Good Data B. Computing Capability C. Higher number of influencing Factors D. Dynamic Environment and Need to Decide Fast E. Need to establish insights from unknown & unexplored factors
  • 21.
    21 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. OSEMN Methodology Overview Public/Web data Cloud Database Centralized database Local machines Obtain Scrub Explore Model iNterpret Cleaning data • Joining data • Standardizing • Missing Values & Duplicates • Data Formatting • Feature Scaling • Data Encoding Exploratory Analysis • Descriptive Statistics • Graphical Analysis • Comparing Groups • Feature Engineering Regression Decision Tree Logistic Regression Cluster Analysis Naive Bayes classifier Data Interpretation Hyperparameter Tuning Iterations Delivering Business Value Soft Skills • Stakeholder Analysis • Communication Plan • Questioning Techniques • Change Management Case Studies
  • 22.
    23 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. BIG Data needs special attention The term ‘Big Data’ is: Ø Frequently abused….. Ø Thrown around by bloggers and journalists to increase hits…. Ø Data ‘too big to fit in Excel’ – Really???
  • 23.
    24 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. What is Big Data: 5 Vs Variety – different types of data Velocity – speed at which new data is generated Volume – amount of data Veracity – lack of quality and accuracy VALUE – turn data into information, into knowledge, into wisdom
  • 24.
    25 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Where does all this Big Data come from? What's important for me, for my organization, my customer? • Immediate • Short Term • Mid Term • Long Term
  • 25.
    26 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Supervised and Unsupervised Learning Unsupervised Supervised
  • 26.
    27 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Supervised Learning Characteristics of Employees Who Buy Our Product Yes - Buy No - Do Not Buy
  • 27.
    28 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Types of Learning - Supervised • Training data includes desired outputs. • Machine is trained on past data and its output to learn the hidden pattern. Hence Supervised
  • 28.
    30 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Unsupervised Learning Segment Employee with known Characteristics Oval Body Oval Body & Square Head Oval Body & Circular Head Oval & Gray Body with Circular Head Oval & White Body with Circular Head
  • 29.
    31 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Types of Learning - Unsupervised • Training data does not include desired outputs • Machine is not trained on any specific output. Hence, the name Unsupervised Learning
  • 30.
    33 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Types of Data Ø Variable or Continuous § Salary in Dollars, Number of Years of experience Ø Discreate or Attribute § Binary - Yes or No, Good or Bad, Present or Not Present § Ordinal - No of Family Members § Categorical – ZIP Code, Education, Profession
  • 31.
    34 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Machine Learning Unsupervised Learning Reinforcement LearningSupervised Learning Target Not Available Target Continuous Data Target Categorical Data Target Categorical Data Target Not Available Regression Classification Clustering Association Classification Control Use Cases: • Ad Popularity • Predicting Prices • Growth Prediction Tabular Data: Audio / Video Data: Use Cases: • Predicting Churn • Fraud Detection • User Conversion Tabular Data: Audio / Video Data: Use Cases: • Segmentation • Recommender Systems Tabular Data: Audio / Video Data: Use Cases: • Market basket analysis Tabular Data: Audio / Video Data: Use Cases: • Real-time Decisions • Gameplay • Learning Skills Tabular Data: Audio / Video Data: Use Cases: • Self Driving Cars • Fault Detection • Robot Navigation Tabular Data: Audio / Video Data:
  • 32.
    35 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Background of the Project Probability of taking Personal Loan q A retail bank is interested in finding out the influence of Cross Selling Efforts to the existing customers w.r.t. decision to take a Personal Loan q The data were collected on various influencing factors given in next slide
  • 33.
    36 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Data Structure 1 of 2 ü Personal Loan - Did this customer accept the personal loan offered in the last campaign? ü Customer ID ü Customer's age in completed years ü Experience #years of professional experience ü Income Annual income of the customer ($000) ü ZIPCode Home Address ZIP code. ü Family size of the customer ü Credit Card Average spending on credit cards per month ü Education Level – Non-Graduate, Graduate; Professional
  • 34.
    37 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Data Structure 2 of 2 ü Mortgage Value of house mortgage if any. ($000) ü Securities Account Does the customer have a securities account with the bank? ü Certificate of Deposit (CD) account with the bank? ü Online Does the customer use internet banking facilities? ü Credit Card – Yes or No ü Does the customer use a credit card issued by UniversalBank?
  • 35.
    38 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Big Data with Large Number of Features is Often Unbalanced Number of Customers by ZIP Code & Education Education ZIP Codes 1 2 3 Grand Total 9307 1 1 90007 5 1 6 90009 4 2 2 8 90011 1 1 1 3 90016 2 2 90018 2 2 4 90019 1 2 2 5 90024 20 12 18 50 90025 5 7 7 19 90027 2 1 3 90028 4 2 2 8 90029 4 1 5 90032 5 2 1 8 90033 4 2 3 9 90034 12 2 6 20 90035 1 1 4 6 90036 5 1 1 7 90037 2 1 2 5 90041 5 1 4 10 90044 1 1 2 90045 3 3 Number of Customers 1. Missing Not at Random (MNAR) 2. Missing at Random (MAR) 3. Missing completely at Random (MCAR) 4. Missing at Random (MAR) Missing completely at Random (MCAR)
  • 36.
    39 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works.
  • 37.
    40 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Ø Missing Values Ø Duplicate Value Ø Dimensionality/Scale Ø Outliers/Extreme Values Ø Combining Data from Multiple Sources SCRUB
  • 38.
    41 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. EXPLORE : How reliable is Analysis on Pivot Table…. No of Family Member % Loan Availed No Yes 1 93% 7% 2 92% 8% 3 87% 13% 4 89% 11% Grand Total 90% 10% Education % Loan Availed No Yes Non Graduate 96% 4% Graduate 87% 13% Post Graduate 86% 14% Grand Total 90% 10% % Avail Loan Family Size Education 1 2 3 4 Overall Non Graduate 1% 1% 11% 10% 4% Graduate 12% 19% 11% 11% 13% Post Graduate 12% 14% 18% 12% 14% Overall 7% 8% 13% 11% 10%
  • 39.
    42 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. EXPLORE - Data Distributions
  • 40.
    43 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Complex relationships between Output and Inputs Number of relationships cannot be fitted with Liner, Quadratic or even advance polynomial equations
  • 41.
    44 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. MODEL - CART
  • 42.
    45 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Cost of Misclassification
  • 43.
    46 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Relative Importance of Factors
  • 44.
    47 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Decision Tree Overall Hit Rate 9.5% Salary < 92.5K 0.7% Income>92.5% 33% Non Graduates 10% Graduates + 72% Family 1,2 Near 0% Family 3+ 38% Focus on Clients with • Income >92.5K • Graduates + • Non-Graduates - Family Size 3+
  • 45.
    48 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Train and Test Model Summary
  • 46.
    49 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Prediction q The model provides ability to predict hit rate for multiple profiles q Individuals with income 150K & Non-Graduates, the hit rate is below 1% if family size is 1 or 2, this goes up to 72% if family size is 3 or 4.
  • 47.
    50 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Confusion – Train v/s Test
  • 48.
    51 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Unsupervised Learning – Cluster Observations q Scenarios where no specific outcome is specified or known q In the same banking environment we have information of 5000 customers profiles. q Objective is to Segment customers in the logical groups
  • 49.
    52 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Cluster Observations - Dendrogram
  • 50.
    53 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Cluster Centroids Variable Cluster1 Cluster2 Cluster3 Cluster4 Cluster5 Grand centroid Age 43.4 46.1 46.7 42.1 45.7 45.3 Experience 18.1 21.0 21.4 17.0 20.7 20.1 Income 29.3 108.2 66.1 188.6 149.6 73.8 Family 2.5 2.1 2.6 2.0 2.0 2.4 CCAvg 0.9 2.6 1.7 4.7 4.0 1.9 Education 2.0 1.6 2.0 1.7 1.5 1.9 Cluster Properties – Income and Credit Card Spend No of Observations. 1998 499 1761 104 158 5000 Chances of Taking Loan. 0%. 29%. 3%. 51%. 43%
  • 51.
  • 52.
    55 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Customer Sentiment Analysis
  • 53.
    56 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Tracing LinkedIn Profile Application ü Swipe in swipe out ü New Employee Onboarding ü Guest/Visitor Management
  • 54.
    57 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Security Enhancement
  • 55.
    58 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. SPACE CUSTOMER FEEDBACK – RESORT BEACH ROOM Far Crowd Water Room Service Fun Great Shower Staff Clean
  • 56.
    59 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. IShallNever MixWater Prof Juran’s Example on Root Cause Analysis
  • 57.
    60 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Threat to Causal Inferences 1. Selection 2. Simultaneity 3. Omitted Variable 4. Measurement Error
  • 58.
    61 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Selection – Sample is Not Representative
  • 59.
    62 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Correlation Causation
  • 60.
    63 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Simultaneity – X and Y Cause Each Other
  • 61.
    64 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Point to Remember q Any such analysis should be used with complete care as misinterpretations can add to frustration and result in disbelief in scientific methods q While we are using advanced analytics tools to further strengthen our ability to identify root cause, Team Engagement, Brainstorming, FMEA, Why-Why Analysis, 6-3-5, Mistake Proofing, Daily Work Management remain equally important q Good old Rule applies Business Problem – Statistical Problem – Statistical Solution – Business Solution
  • 62.
    65 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Data Enrichment - Using Secondary Data Data enrichment is the process of creating new features by introducing data from external sources. ü Information on Bank’s Customer Salary Increase Decrease can be extracted from Saving Bank Transaction Records ü Expenditure pattern can be extracted from Credit Card Bills ü Addition in family member can be accessed through Birth and Death Register, Mediclaim’s, etc.
  • 63.
    66 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Opportunity to Use AI and ML q The model can be linked to multiple data sources within/outside bank to update the information on existing parameters and add additional parameters like Age of Family Members, Changes in Income and investment, etc q This information can be used to auto trigger the agents to enable enhancement of cross selling effectiveness, launching new products etc.
  • 64.
    Licensing Open SourceOpen Source SaaS High / One-Time SaaS Ease of use Difficult (Requires Coding) Difficult (Requires Coding) Difficult (Requires Coding) Medium Easy Applications Broad Broad Broad Narrow Dataset Specific Easy to Understand Difficult (Requires Coding) Difficult (Requires Coding) Difficult (Requires Coding) Medium Easy Setup Time Months Months Weeks Weeks 10 mins Training Speed Weeks / Months Weeks / Months Days Days < 1 minute Deployment Time Months Months Weeks Weeks < 1 minute Engineer Involvement High (Requires Coding) High (Requires Coding) High (Requires Coding) Medium None Cloud Specific? No No Yes (Only Supports AWS) No No Support N/A N/A via Email via Email 24x5 Hands-On Support Software Landscape & Comparison
  • 65.
    68 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. BMGI Certified Business Analytics Professional Module 1 : Foundation Module 2 : Advanced Modelling + Python & R (Optional) Module 3 : Forecasting & Visualisation Module 4 - Optimisation, AI & ML
  • 66.
    69 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Case Studies
  • 67.
    70 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Sample Case Studies 1. Reduction in Processing Time and Error Rate 2. Profitability improvement in multi channel Sales 3. Predicting and reducing days of Sales Outstanding 4. Deriving actionable insights from Employee Satisfaction Survey 5. Reduction in Customer Returns and Complaints 6. Reduction in Procure to Pay Cycle Time
  • 68.
    71 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works.
  • 69.
    72 © BMGI. Exceptas may be expressly authorized by a written license agreement signed by BMGI, no portion may be altered, rewritten, edited, modified or used to create any derivative works. Wish you All the Best!!!!