Data Analytics - IRAS Times

Data Analytics - IRAS Times

Data Analytics National Academy of Indian Railways Workshop on 'New Financial Initiatives' in Indian Railways For Principal/ Coordinating FA&CAOs 23 December 2015 There are Lies Damned lies; and? Statistics 23-Dec-15 Data Analytics - NAIR 2 Data Analytics What do you understand by it? 23-Dec-15 Data Analytics - NAIR 3 Plan for the session An example Data Analytics

Digital Dashboard Use of Numbers Data on Indian Railways Reorganising Statistical Units Analytics Ecosystem on IR A word of caution 23-Dec-15 Data Analytics - NAIR 4 An example from Railways Not Financial Data 23-Dec-15 Data Analytics - NAIR 5 An illustration Accidents on Indian Railways 23-Dec-15 Data Analytics - NAIR 6

Cause wise Analysis of Consequential Train Accidents over IR (2006-07 to 2014-15) 2006-07 2007-08 2008-09 2009-10 2010-11 2011-12 2012-13 2013-14 2014-15 Cause Failure of Railway 84 87 83 63 56 52 46 51 62 staff Failure of other 87 81 68 75 57 63 59 56 58 than Railway staff Failure of 7 9

0 6 5 5 6 4 2 equipment Sabotage Combination of factors 8 1 7 0 13 4 14 1 16 3 6 1

3 0 3 0 3 0 Incidental Could not establish 7 1 8 2 5 4 4 2 4 0 3 1

7 0 3 0 8 0 None Held Awaited Total 0 0 195 0 0 194 0 0 177 0 0 165

0 0 141 0 0 131 1 0 122 0 1 118 2 0 135 23-Dec-15 Data Analytics - NAIR 7 Accidents on Indian Railways 195

194 177 165 141 135 131 122 2006-07 23-Dec-15 2007-08 2008-09 2009-10 2010-11 Data Analytics - NAIR 2011-12 2012-13 118

2013-14 2014-15 8 Accidents on Indian Railways 195 194 177 165 141 135 131 122 2006-07 23-Dec-15 2007-08 2008-09

2009-10 2010-11 Data Analytics - NAIR 2011-12 2012-13 118 2013-14 2014-15 9 Accidents on Indian Railways 195 194 177 165 141 135

131 122 2006-07 23-Dec-15 2007-08 2008-09 2009-10 2010-11 Data Analytics - NAIR 2011-12 2012-13 118 2013-14 2014-15 10 Consequential Accidents over the years

Collisions Manned Level Crossing Accidents 100 96 Derailments Miscellaneous Accidents Fire Unmanned Level Crossing Accidents 85 80 80 72 65 65 62 63 55 54

8 4 7 8 2006-07 23-Dec-15 8 13 12 5 4 2007-08 3 7 7 2008-09 9 2

5 4 2009-10 5 2 5 9 4 1 2010-11 53 49 48 7 2 2011-12 Data Analytics - NAIR

6 53 50 47 9 5 2012-130 4 7 4 3 2013-14 5 6 6 5 2014-15 11

Accidents by type in 2006-2015 Unmanned Level Crossing Accidents; 37.45% Manned Level Crossing Accidents; 4.21% Miscellaneous Accidents; 2.47% Fire; 3.05% Collisions; 4.86% Derailment; 47.97% 23-Dec-15 Data Analytics - NAIR 12 Consequential Accidents over the years 72 Unmanned Level Crossing Accidents 65 Derailments

65 96 62 54 48 100 85 53 50 80 80 47 63 55 49 23-Dec-15 Data Analytics - NAIR 53

13 Consequential Accidents over the years 72 Unmanned Level Crossing Accidents 65 Derailments 65 96 62 54 48 100 85 53 50 80 80 47

63 55 49 23-Dec-15 Data Analytics - NAIR 53 14 Seasonal variations? A widely prevalent belief Specific types of accidents have higher frequency in different times of the year 23-Dec-15 Data Analytics - NAIR 15 136 131 120 April

125 115 109 102 119 117 103 103 98 Month-wise distribution of Total Accidents in the period 2006-2015 May 23-Dec-15 June July August September October November December January Data Analytics - NAIR

February March 16 136 131 120 April 125 115 109 102 119 117 103 103 98 Month-wise distribution of Total Accidents in the period 2006-2015

May 23-Dec-15 June July August September October November December January Data Analytics - NAIR February March 17 136 131 120 April 125 115 109 102

119 117 103 103 98 Month-wise distribution of Total Accidents in the period 2006-2015 May 23-Dec-15 June July August September October November December January Data Analytics - NAIR February March 18 136

131 125 120 117 January 119 115 109 103 103 102 98 Month-wise distribution of Total Accidents in the period 2006-2015 February 23-Dec-15 March April

May June July Data Analytics - NAIR August September October November December 19 136 131 125 119 120 117

115 109 103 102 103 98 Month-wise distribution of Total Accidents in the period 2006-2015 July August 23-Dec-15 September October November December January

Data Analytics - NAIR February March April May June 20 May-June 23-Dec-15 Data Analytics - NAIR 21 July-September 23-Dec-15 Data Analytics - NAIR 22

December - January 23-Dec-15 Data Analytics - NAIR 23 March April October - November 23-Dec-15 Data Analytics - NAIR 24 Month-wise distribution of Different types of Accidents in the period 2006-2015 Unmanned Level Crossing Accidents 65 79 76 52 60 46 57

46 45 45 42 39 54 52 39 54 46 37 52 44 41 33 27

Derailment 23-Dec-15 Data Analytics - NAIR 25 46 Month-wise distribution of Different types of Accidents in the period 2006-2015 Unmanned Level Crossing Accidents 65 79 76 52 60 46 57 46 45 45 42

39 54 52 39 54 46 37 52 44 41 33 27 Derailment 23-Dec-15 Data Analytics - NAIR 26

46 Accidents by type in 2006-2015 Monsoon-Gangetic Plain Miscellaneous Accident ; 1.96% Monsoon-West Unmanned Level Crossing Accidents; 37.45% UMLC Accident ; 47.06% UMLC Accident ; 32.35% MLC Accident ; 7.84% Manned Level Crossing Accidents; 4.21% Fire ; 2.94% Collision ; 2.94% Miscellaneous Accidents; 2.47%

Fire; 3.05% MLC Accident ; 2.94% Fire ; 5.88% Collision ; 2.94% Collisions; 4.86% Derailment ; 41.18% Derailment ; 51.96% Derailment; 47.97% Fog-Gangetic Plain Harvest Seasons UMLC Accident ; 36.89% Miscellaneous Accident ; 2.91% MLC Accident ; 10.68% UMLC Accidents; 36.55%

Monsoon-East UMLC Accident ; 22.22% Fire ; 5.56% Miscellaneous Accidents; 1.84% MLC Accident; 4.14% Collision ; 7.77% Collision; 4.37% Fire; 3.68% Fire ; 2.91% Derailment ; 38.83% 23-Dec-15 Derailment; 49.43% Data Analytics - NAIR Derailment ; 72.22% 27

Causes of Accidents 2006-15 Combination of factors; None Held; 0.06% Could not be estb.; 0.85% Incidental; 4.11% 0.56% 4.22% Sabotage; Failure of equipment ; 4.72% Failure of Railway staff; 49.97% Failure of other than Railway staff; 35.51% 23-Dec-15 Data Analytics - NAIR 28

Causes of Accidents 2006-2015: Number wise break-up 04 11 4 19 None Held 33 05 20 0 14 Could not be estb. Incidental 24 02 15 2 10 18 109 103

Combination of factors 02 17 2 18 18 Sabotage Failure of equipment 118 107 293 2 16 14 14 3 11 06 8 78 86

Failure of other than Railway staff 249 186 17 18 9 18 07 9 84 81 23-Dec-15 3 00 -2 2 0 20 241 14 6

76 4 00 -2 3 0 20 5 00 -2 4 0 20 043 16 5 131 65 57 63 56

52 75 17 036 0403 3 59 57 46 51 218 033 58 120 88 85 2 00 -2

1 0 20 45 4 13 0 161 119 1 00 -2 0 0 20 Failure of Railway staff 6 00 -2 5 0 20 7 00

-2 6 0 20 8 00 -2 7 0 20 75 09 20 08 20 Data Analytics - NAIR 63 09 20 10 20 11

20 10 20 12 20 11 20 3 01 -2 2 1 20 4 01 -2 3 1 20 60 5 01 -2 4 1

20 29 Causes of Accidents 2006-2015: Percentage break-up None Held 4 11 4 19 33 Could not be estb. 05 02 15 20 2 0 10 14 18 24 Incidental 02 17 2 18 18

Combination of factors 2 3 11 16 06 14 8 14 Sabotage 1 7 1 8 Failure of equipment 1 8 0 7 9 9 Failure of other than Railway staff 0 1 1

4 3 1 7 3 6 03 5 16 6 2 4 1 4 5 4 14 13 0 6 Failure of Railway staff 0 21 4

03 8 3 03 3 5 109 103 118 293 57 81 84 76 63 75 57 63

56 58 59 249 186 01 -0 20 0 20 86 78 107 2 00 -1 2 0 20 23-Dec-15 3

00 -2 2 0 20 161 4 00 -3 2 0 20 119 120 88 85 05 -4 20 0 20 06 -5 20 0 20

07 -6 20 0 20 08 -7 20 0 20 75 9 00 2 08 20 Data Analytics - NAIR 09 0 2 10 20 10 20

1 01 2 - 52 11 20 2 01 2 - 13 -2 20 1 20 60 51 46 14 -3 20 1

20 15 -4 20 1 20 30 Causes of Accidents 2006-2015: Trends Failure of Railway staff Failure of other than Railway staff 62% 60% 48% 48% 48% 53% 50% 45% 51% 51% 43% 45% 44% 43%

42% 44% 43% 43% 40% 37% 40% 40% 38% 42% 34% 33% 33% 38% 25% 23% 23-Dec-15 Data Analytics - NAIR 31

Causes of Accidents 2006-2015: Trends Failure of Railway staff Failure of other than Railway staff 62% 60% 48% 48% 48% 53% 50% 45% 51% 51% 43% 45% 44% 43% 42% 44% 43% 43% 40%

37% 40% 40% 38% 42% 34% 33% 33% 38% 25% 23% 23-Dec-15 Data Analytics - NAIR 32 Derailments Cause-wise Analysis (2006-07 to 2014-15) Sabotage; 10.74% Combination of factors; 0.91% Could not estb.; 1.06%

Failure of equipment ; 6.66% Failure of other than Railway staff; 6.51% Incidental; 6.81% None Held; 0.30% Awaited; 0.15% Failure of Railway staff; 66.87% 23-Dec-15 Data Analytics - NAIR 33 Manned LC Accidents Cause wise Analysis (2006-07 to 2014-15) Failure of other than Railway staff; 18.97% Combination of factors; 1.72% Failure of Railway staff; 79.31%

23-Dec-15 Data Analytics - NAIR 34 Unmanned LC Accidents Cause wise Analysis (2006-07 to 2014-15) Failure of Railway Staff; 1.00% Failure of other than Railway Staff; 99.00% 23-Dec-15 Data Analytics - NAIR 35 Preventive Measures To Curb UMLCs As on 01.04.2015, there were approximately 29487 LC on IR out of which 19047 (64.6%) are Manned and 10440 (35.4%) are Unmanned. Progress made in elimination of LC in last 5 years & up to July, 2015 by Closure, Merger, Subway and Manning are as under UMLC By Closure/ Merger/Subway

By Manning Total 2010-11 2011-12 2012-13 2013-14 2014-15 Total 2010-15 2015-16 (Up to July15) 800 481 700 777 721 3479 206 434 777 463 325 427 2426 65 1234

1258 1163 1102 1148 5905 271 MLC 2010-11 2011-12 2012-13 2013-14 2014-15 By Closure 133 225 257 301 310

23-Dec-15 Data Analytics - NAIR Total 2010-15 2015-16 (Up to July15) 1226 89 36 Data Analytics 23-Dec-15 Data Analytics - NAIR 37 What is data analytics? Derive meaning from data by incorporating Statistics Mining of data Visualisation Extracting actionable data in a manner that supports decision-making 23-Dec-15

Data Analytics - NAIR 38 What is data analytics? Massive expansion in the ability of computers to handle data leading to certain other crucial items now becoming possible: Machine Learning Database engineering All this to solve complex problems 23-Dec-15 Data Analytics - NAIR 39 What is data analytics? Caters to Big Data with a view to capturing major and seemingly minor relationships of performance indices Caters to day to day reporting needs Caters to ad hoc querying Provides analytical dashboards and alerts Provides comprehensive information and actionable insights for taking informed decisions 23-Dec-15

Data Analytics - NAIR 40 What has changed? More powerful computers Methods to obtain data directly from source Data feeds available from diverse sources Algorithms to extract information from unstructured data 23-Dec-15 Data Analytics - NAIR 41 4 Vs of Big Data 23-Dec-15 Data Analytics - NAIR 42 Insights into the data Traditional business intelligence What happened?

Diagnostic analytics Why is it happening? Predictive What will happen in future? Prescriptive What should we do? 23-Dec-15 Data Analytics - NAIR 43 Communication Flow in Analytics Analytics Team Functional User Functional Analyst providing liaison between User and Technology Team Technology team Software Programmers to extract data from databases and prepare it for analytical models Data Scientists to decide choice of model and provide interpretation of analytical output in Statistical Analysts/Econometricians for developing appropriate logical and physical data models Quantico Analysis Team testing the quality of product from non-functional point of view

functional terms Present Action? Delays in access to data Limited electronic filing and diary Limited integration with policy and financials Multiple manual processes to enter, correct, extract and analyze data Delayed decision making based on limited information and insights 23-Dec-15 Data Analytics - NAIR 46 Possibilities in future Action Shorter duration for analysis Shorter time-lapse for decision Quality of data analysis provides decision support Continuous feedback on decisions taken Rapid course correction if needed Easier launching of new initiatives 23-Dec-15

Data Analytics - NAIR 47 Features determining analytics maturity Use of real time data Electronic filing and diary in a centralised model for each business Information integrated across the organization Advanced modeling techniques used to evaluate across functional areas Fully simulated business operations to evaluate decisions A crucial tool Digital Dashboard 23-Dec-15 Data Analytics - NAIR 49 23-Dec-15 Data Analytics - NAIR 50 23-Dec-15

Data Analytics - NAIR 51 23-Dec-15 Data Analytics - NAIR 52 The Digital Dashboard Visual depiction Timely alerts 23-Dec-15 Multi device Drilldown capability Data Analytics - NAIR 53 The Digital Dashboard Relevance Right insight at the right time Convenience Readily available when needed

Validation Checked for errors and model validity 23-Dec-15 Data Analytics - NAIR 54 23-Dec-15 Data Analytics - NAIR 55 Statisticians delight Use of numbers 23-Dec-15 Data Analytics - NAIR 56 la loi des grands nombres (Law of Large Numbers) If the expected result of an experiment is random

And the experiment is repeated a large number of times Then the results tend to stabilise over a period of time 23-Dec-15 Data Analytics - NAIR 57 23-Dec-15 Data Analytics - NAIR 58 23-Dec-15 Data Analytics - NAIR 59 Statisticians Delight 23-Dec-15 Data Analytics - NAIR 60 Dependent and Independent variables

y = a0x0+ a1x1+ a2x2+ a3x3+ 23-Dec-15 Data Analytics - NAIR 61 Dependent and Independent variables Throughput = f (Availability of power reliable signalling, Crew, TXR cleared RS, efficient controllers, motivated staff and officers, reliable equipment of all types, etc.) 23-Dec-15 Data Analytics - NAIR 62 Optimisation Max. [y (ax+ by+ cz)] 23-Dec-15

Data Analytics - NAIR 63 Optimisation Maximise Throughput Subject to: loco failures, signal failures, crew non-availability, failing rolling stock, absence of crucial staff, demotivated staff and officers, other asset failures, etc. 23-Dec-15 Data Analytics - NAIR 64 Our position Data on Indian Railways 23-Dec-15 Data Analytics - NAIR 65 Recording of data on IR I

Digitally Captured Digitally Reported II Manually Captured Digitally Reported IV Digitally Captured Manually Reported III Manually Captured Manually Reported Quadrant I is ideal. However, on IR data can be available to varying degrees in all Quadrants. 23-Dec-15 Data Analytics - NAIR 66 Data on Indian Railways Several IT projects on IR have potential for data analytics: Data Warehouse for PRS Software for Locomotive Asset Management (SLAM) for Electric Loco Sheds

Data Warehouse for UTS Track Management System (TMS) Data Warehouse for FOIS MIS for Land & Amenities on IR Control Office application Integrated Coach Monitoring System (ICMS) Loco Sheds Management System (LSMS) Traction Distribution Management System (TDMS) Signaling Maintenance Management System (SMMS) Expenditure side application of PRIME/AFRES/IPAS/ e-Recon/ARPAN etc. 23-Dec-15 Data Analytics - NAIR 67 Data on Indian Railways Tickets Maintenance

reserved / unreserved P Way and fixed assets Freight Maintenance RRs Rolling stock Train movement Materials control office purchase and depots Crew HR movement and lobbies manpower deployment and staff welfare Costing and accounting all of the above 23-Dec-15

Data Analytics - NAIR 68 Change Reorganising Statistical Units 23-Dec-15 Data Analytics - NAIR 69 Reorganizing Statistical Unit to Analytics Unit Dynamic officers with adequate field experience and IT knowledge should lead the team Identify comparatively younger staff and train them in data handling and analysis Involve young JE level staff of EDP centres as part of the Unit Unit should provide analyzed inputs to all departments 23-Dec-15 Data Analytics - NAIR 70 Something new

Analytics Ecosystem on IR 23-Dec-15 Data Analytics - NAIR 71 An Analytics Ecosystem for IR 23-Dec-15 Data Analytics - NAIR 72 Possible Areas for Analytics Predictive maintenance Dynamic pricing Evaluating different marketing strategies Improving capacity utilization/route congestion Real time management of linear and rolling stock assets 23-Dec-15 Data Analytics - NAIR 73

Possible Areas for Analytics Expenditure on Fuel consumption Correlation with changes in fuel prices, Specific Fuel Consumption of locomotives, route electrification and transport output RCD wise energy rate and consumption data Automatic alerts in case the inventory of fuel increases to more than 10 days Expenditure on Traction Bills Correlation with prices of traction, Loco-wise and EMU coach wise energy consumption and transport output TSS-wise energy rate and consumption data Automatic alerts in cases of diesel locos running under wires 23-Dec-15 Data Analytics - NAIR 74 Some theoretical stuff Dont rely too much on predictive capacity of current data 23-Dec-15 Data Analytics - NAIR 75 An economist can affect the economy as much as

The weatherman the weather 23-Dec-15 Data Analytics - NAIR 76 Logical fallacy Post hoc ergo propter hoc? The rooster crows before sunrise. Ergo the rooster causes sunrise. Cum hoc ergo propter hoc? Rate of deaths in India due to TB increased even as civilian deaths during war in Iraq was increasing Ergo War in Iraq was the cause of increase in TB death rate in India 23-Dec-15 Data Analytics - NAIR 77 Chaos Theory Chaos When the present determines the future, but the approximate present does not approximately determine the future

23-Dec-15 Data Analytics - NAIR 78 Chaos Theory Does the flap of a butterflys wings in Brazil set off a tornado in Texas? The butterfly does not power or directly create the tornado The flapping of wings by the butterfly is a set of initial conditions which are followed by the tornado the final result Had the butterfly not flapped its wings, the trajectory of the system might have been vastly different For want of a nail the shoe was lost. For want of a shoe the horse was lost. For want of a horse the rider was lost. For want of a rider the message was lost. For want of a message the battle was lost. For want of a battle the kingdom was lost. And all for the want of a horseshoe nail. 23-Dec-15 Data Analytics - NAIR 79 Thank You 23-Dec-15 Data Analytics - NAIR

80

Recently Viewed Presentations

  • Lender Underwriter Training Continental Breakfast Sponsored by: Oppenheimer

    Lender Underwriter Training Continental Breakfast Sponsored by: Oppenheimer

    freddie r soto gloria villarreal jason w hunt john j bruce john m hartung joseph s tuczak karen m safer lavona g llewellyn linda monger michael glasgow mikki webb peter c hazen regina a casabal shirley a caruso suechen smith...
  • Resource Management with Microsoft Project and Project Server/Project

    Resource Management with Microsoft Project and Project Server/Project

    Microsoft Project Management Solution Spectrum. Advanced TeamCollaboration . Enterprise . Project & Portfolio. Management. Project Pro sync with SharePoint . Project Pro for Office365. Low. Medium. High. Level of Process. Project Server . or Project Online (Built on SharePoint ®...
  • HW-147T: Building high quality Windows PCs using the ...

    HW-147T: Building high quality Windows PCs using the ...

    Building high quality Windows PCs using the assessment and deployment kit. Michael Beck. Principal Group Program Manager. Microsoft Corporation. HW-147T. 9/16/2011 2:00 PM ... Building high quality Windows PCs using the assessment and deployment kit
  • SciENcv : a brief introduction

    SciENcv : a brief introduction

    Integration of my NCBI Tools. Method A&B. Method C&D. Information from the fields in PubMed, PubMed Central and the NIH Manuscript Submission System are all part of NCBI and transferred to other NCBI fields.
  • Mihail Alexandrovici Bakunin (30 mai 1814 - 1

    Mihail Alexandrovici Bakunin (30 mai 1814 - 1

    REALIZATOR-HOSSU BOGDAN,Colegiul Tehnic "George Bariţiu", Baia Mare; Profesor-coordonator-director-adjunct Vasile Pop REALIZATOR-HOSSU BOGDAN,Colegiul Tehnic "George Bariţ
  • Aim: What the Ming Dynasty all about?

    Aim: What the Ming Dynasty all about?

    Verdana Arial Wingdings Times New Roman Globe 1_Globe Ming and Qing Dynasties Slide 2 Slide 3 The Rise of a Dynasty A Return to Confucianism An Age of Exploration Slide 7 Slide 8 Slide 9 Slide 10 Ming Vase Ming...
  • Being Kinder to Myself Elaine Beaumont, Lecturer University

    Being Kinder to Myself Elaine Beaumont, Lecturer University

    Learning to observe and monitor physical, emotional, behavioural and cognitive reactions using thought records/charts/cost benefit analysis. Cognitive restructuring to examine core beliefs, dysfunctional assumptions and negative automatic thoughts. Compassionate letter writing - focusing on being kind, supportive and nurturing to...
  • Term 1  Sept  Dec Year 9 Introduction to

    Term 1 Sept Dec Year 9 Introduction to

    Physical Theatre (Movement, Physicality & Theatre Style) Year 10. Component 1 (40%) Devising (Introduction to Brecht) Component 1 - Mock (40%) Devising (Portfolios and Evaluation) Year 11. Component 1 (40%) Devising - Creating Performance (end of performance) Component 1 (40%)...