Skip to content

Risk Academy, led by the Nexus Consortium · Risk Science and Methods · Department of Risk Management

Models that are wrong or misused, and the controls that manage them

মডেল ঝুঁকি ও যাচাই

A model is a decision tool with an error term, and model risk is what happens when the error is ignored.

Map model risk and validation on a real placeSee the courses

Illustrative: a joint probability density whose dependence shifts over time

01পরিধি

What the discipline covers

The subtopics its courses and labs teach.

কোর্স
In preparation
স্তর
L3 Advanced
গবেষণাগার
In development
  • Model risk governance
  • Model validation
  • Model monitoring and drift
  • Explainability and fairness
  • Third-party models
  • Generative AI in finance
  • Human-in-the-loop controls
  • Model inventories

02মানচিত্র

Build a twin of any place

Search any place or open the world; the layers and generators follow model risk and validation on the planet map, the globe and in 3D.

The central Apennines: L'Aquila, Amatrice and Norcia: a flat plan of the faculty twin. Terrain 24 to 2,805 m; 3 reference earthquakes, M6.2 to M6.6; Earthquakes M4.5 and above, past week: read live when the 3D twin opens; USGS ShakeMap intensity contours: read live when the 3D twin opens. Contours every 500 metres from the elevation grid; reference earthquakes as marks.
The central Apennines: L'Aquila, Amatrice and Norcia, Italy. Flat plan of the same data: relief, the three reference epicentres with magnitude and depth, and the numbers below; the sampler's figure shows one precomputed field with its parameters. The 3D twin opens on devices that can draw it, once its data and code have been checked.
  • Terrain 24 to 2,805 m
  • 3 reference earthquakes, M6.2 to M6.6
  • Earthquakes M4.5 and above, past week: read live when the 3D twin opens
  • USGS ShakeMap intensity contours: read live when the 3D twin opens
Credits and dates per layer are listed under the stage. Relief in the twin is exaggerated ×1.5 for legibility.

Drag to move, right drag or Shift with a drag to turn and tilt, Ctrl with the wheel to zoom. Keyboard: arrows move, Shift with arrows turn and tilt, + and – zoom, 0 resets, V steps through the views, Escape clears.

Flat plan of the same data: relief, the three reference epicentres with magnitude and depth, and the numbers below; the sampler's figure shows one precomputed field with its parameters. The 3D twin opens on devices that can draw it, once its data and code have been checked.

The boundaries, names and designations shown do not imply any judgement on the legal status of any territory, or any endorsement or acceptance of such boundaries. Countries: Natural Earth (public domain).

03The subject

The subject, methods and measures and debates and bodies

The discipline

The field and its scope

For model validators, model owners, data scientists and risk committees who must decide whether a model is fit for the decision it supports and keep it that way in use.

Model risk and validation study the losses and harms that follow when decisions rest on models that are wrong, misused or no longer fit the world, and the controls that reduce that risk: inventories, tiering by materiality, independent validation, performance monitoring, change control and clear limits on use. The field took shape in banking after pricing and credit model failures and was codified in the Federal Reserve's SR 11-7 (2011) and the Bank of England's SS1/23 (2023); its methods now apply to insurance, catastrophe, climate, epidemiological and machine learning models, including generative systems whose outputs are hard to test against a ground truth.

The discipline is organised around the model life cycle. Development documents purpose, data, assumptions and limitations; validation tests conceptual soundness, implementation and outcomes with back-testing, benchmarking and sensitivity analysis; monitoring tracks drift in inputs and performance; governance records the inventory, the tiering and who may approve what. Explainability and fairness testing enter where models affect people. The standing questions are what validation can show for models of rare events, how much independence is enough, and how to govern third-party and foundation models whose internals are not available to the user.

Measures

Quantities and units

QuantityUnit or scaleWhat to know
Back-test exceedance rateCount of exceptions per period against the expected count at the model's confidence levelFor a quantile forecast such as value at risk, the number of times loss exceeded the forecast. Tests such as Kupiec's are weak for small samples: a year of daily data cannot reject a 99 per cent model whose true rate is 1.5 per cent.
Discrimination (AUC, Gini) and calibration (expected versus observed rate)Dimensionless, 0.5 to 1 for AUC; ratio or calibration slope for calibrationStandard performance statistics for classification models. Both are estimated from a sample with a confidence interval and both degrade when the population shifts; discrimination can stay high while calibration fails.
Population stability index and feature drift statisticsDimensionless index; or a test statistic per featureMeasures of how far the input distribution has moved from the development sample. Thresholds in use are conventions, not derived from loss consequences.
Model materiality tierOrdinal tier from exposure, complexity and useDrives the depth and frequency of validation. It is a governance judgement; the uncertainty lies in whether exposure through a model is known at all, especially for models embedded in other models.
Sensitivity of output to assumptionsChange in output per unit change in a parameter, or a Sobol index share of varianceShows which assumptions matter. Local sensitivities miss interactions; global sensitivity analysis needs many model runs that slow models cannot afford.
Validation finding count by severity and ageCounts; days openThe operational state of the control. A finding backlog with long ages signals that validation has no teeth; the count alone says nothing about the risk the findings carry.

পদ্ধতি

Methods and their limits

Each method is listed with what it assumes and where it fails.

পদ্ধতিAssumesFails when
Independent validation (conceptual soundness, implementation testing, outcomes analysis)That a team outside the model's development line can read the documentation, reproduce key results, test the code and judge the assumptions against the use.When validators lack the domain knowledge or the data to reproduce, when the model is a black box from a vendor, or when validation becomes a document review that never runs the model.
Back-testing and out-of-sample testingThat past outcomes are a fair test of the model's future use, that enough independent observations exist, and that the test is set before the results are seen.For rare events where the test has no power, when the data were used in development, or when the environment changed so that past performance no longer bears on the future.
Benchmarking and challenger modelsThat a simpler or alternative model, run on the same data, exposes where the production model's assumptions drive its output.When the benchmark shares the production model's data and blind spots, or when agreement between models is read as validation against reality.
Monitoring for drift and performance decayThat input distributions and outcomes are observed in use with little delay, and that thresholds trigger review or retraining.When outcomes arrive years later (credit default, climate), when the outcome is itself shaped by the model's decisions, or when thresholds are set without reference to the harm from a stale model.
Explainability and fairness testingThat feature attributions (for example Shapley values) and group performance comparisons reveal whether a model relies on acceptable factors and treats groups comparably.When attributions are unstable across methods, when fairness metrics conflict and no rule chooses between them, or when the protected attribute is not recorded and proxies cannot be traced.
Model inventory with tiering and use restrictionsThat every model can be found, described, assigned an owner and a tier, and that use beyond the documented purpose is detectable.When spreadsheets, embedded rules and third-party components are not counted as models, or when the inventory is a list without links to validation evidence.

Debates

What the field debates

Should high-stakes models be interpretable by design, or is post-hoc explanation of complex models enough?
Rudin (2019) argues that interpretable models perform comparably on most tabular problems and that post-hoc explanations are approximations of an approximation. The other side points to tasks (images, text, weather) where complex models are far more accurate, to supervisory acceptance of explanation methods, and holds that the requirement is reliable testing and monitoring rather than a particular model class.
What does validation mean for models of rare events that cannot be back-tested?
One view, following Oreskes and co-authors (1994), holds that such models can be confirmed in part but never validated, so the control is transparency of assumptions and sensitivity. The other holds that component-level testing, benchmarking and historical event reconstruction give adequate evidence, and that refusing the word validation leaves decision makers without a usable verdict.
Can generative and foundation models be governed under existing model risk frameworks?
Supervisors and SS1/23 hold that the principles apply: define the use, test for it, monitor and restrict. Critics reply that the user cannot inspect training data or weights, that outputs are open-ended so outcome testing is partial, and that NIST's AI Risk Management Framework and ISO/IEC 42001 add the missing elements of use-case governance and incident reporting; the evidence base of documented failures is still young.

Bodies and data

Bodies that set practice, and the data they publish

Bodies and standards are named so learners can find and apply their published work; the Academy teaches their use and links to the publisher, and no affiliation is implied.

BodyRole in the field
Board of Governors of the Federal Reserve System and the Office of the Comptroller of the CurrencyIssued SR 11-7 / OCC 2011-12, the supervisory guidance on model risk management that defined the field for banks.
Bank of England Prudential Regulation AuthorityIssued supervisory statement SS1/23 on model risk management principles for banks, in force from May 2024, extending the scope to all models including those using artificial intelligence.
National Institute of Standards and Technology (NIST)Publishes the AI Risk Management Framework (AI 100-1) and its generative AI profile, used as a reference for governing machine learning models beyond finance.
ISO/IEC JTC 1/SC 42 Artificial intelligenceMaintains ISO/IEC 42001 (AI management systems), ISO/IEC 23894 (AI risk management guidance) and related standards on data quality and bias.
European Banking Authority and European Central BankSet model requirements for internal ratings and run the targeted review of internal models that tests how banks' models are validated.
DatasetSteward
AI Incident DatabaseResponsible AI Collaborative
OECD AI Incidents MonitorOrganisation for Economic Co-operation and Development
Supervisory stress test scenarios and resultsBoard of Governors of the Federal Reserve System

04পদ্ধতি

From a model's claim to a back-test verdict: expected exceptions, observed exceptions and the zone the count falls in

A model that claims a confidence level makes a testable promise about how often outcomes will exceed its figure; counting those exceptions over a test window and comparing the count with the expected distribution gives the verdict, and the verdict feeds the model's tier and its permitted use.

Sources: Board of Governors of the Federal Reserve System, SR 11-7: Guidance on Model Risk Management (2011); Basel Committee on Banking Supervision, Supervisory framework for the use of backtesting in conjunction with the internal models approach (1996); Bank of England PRA, SS1/23 Model risk management principles for banks (2023); Box, Science and statistics, Journal of the American Statistical Association (1976); Kupiec, Techniques for verifying the accuracy of risk measurement models, Journal of Derivatives (1995).

05At a glance

At a glance, with 5 related topics

3D গ্রাফ

ঝুঁকি বিজ্ঞান ও পদ্ধতি অনুষদ

265 nodes1,691 relations9 views

Explore the 3D graph: Faculty of Risk Science and Methods

06কনসোল

How Model risk and validation is taught, as a process you can run

Press Play to run this process from this page's own data.

মডেল ঝুঁকি ও যাচাই

How Model risk and validation is taught

5 subjects in scope, 3 key terms, courses in development, 1 lab, in the Risk Management department of Risk Science and Methods: from the scope to the method and its terms, the courses that teach it, the labs on the twin, the assessment and the record.

  1. 01 Scope

    1. 01

      Model risk governance

      Start · the discipline

      Model risk governance

    2. 02

      Model validation

      Start · the discipline

      Model validation

    3. 03

      Model monitoring and…

      Start · the discipline

      Model monitoring and drift

    4. 04

      Explainability and…

      Start · the discipline

      Explainability and fairness

    5. 05

      Third-party models

      Start · the discipline

      Third-party models

  2. 02 Method and terms

    1. 06

      From a model's claim to…

      Model · the learner

      A model that claims a confidence level makes a testable promise about how often outcomes will exceed its figure; counting those exceptions…

    2. 07

      মডেল ঝুঁকি

      Record · the discipline

      The potential for loss or poor decisions from errors in models or their misuse.

    3. 09

      Validation

      Record · the discipline

      Independent review that a model is sound and fit for its purpose.

    4. 11

      Back-testing

      Record · the discipline

      Comparing model predictions with what happened.

  3. 03 Courses

    1. 08

      RAT-217 Model Risk…

      Record · the learner

      Model Risk Management and Validation

  4. 04 Labs

    1. 10

      টুইন

      Exercise · the learner, on the faculty twin

      The method runs on the faculty twin and the admitted map layers.

  5. 05 Assess

    1. 12

      Module checks

      Gate · the course

      Each module ends with a short check against its outcomes.

    2. 13

      Assessed work

      Decision · the learner

      The assessed work that closes a course: a lab record, a case memo or an assignment, reviewed against the rubric.

  6. 06 Record

    1. 14

      সমাপ্তি নথি

      Outcome · the learner

      The record stacks into the discipline achievement.

    2. 15

      Discipline achievement

      Record · the Academy

      The discipline achievement follows the discipline rule, read from the records.

Model risk and validation · Risk Management An outline of how the work is designed to run; it does not report the status of any decision or record.

07শিক্ষা

Learning, practice and connections

অর্জন

Achievements in this discipline

  1. Course achievements

    One for each course with this as its primary discipline, earned when you pass the course's assessed activities.

  2. Discipline: Model risk and validation

    Complete 3 assessed courses in model risk and validation, at least 2 with it as their primary discipline and at least 1 at Intermediate (L2) or above, totalling at least 45 measured learning hours. In development.

  3. Department: Risk Management

    Earn two discipline achievements in the Department of Risk Management. In development.

  4. Programme: Nexus-Based Risk Management flagship pathway

    Complete every core course of the Nexus-Based Risk Management flagship pathway under assessed completion, ending with its capstone. In development.

An achievement is a record of the learning completed and assessed here. How recognition works.

পথক্রম

Pathways that teach it

Where this is practised

Where the subject is practised

  • The Global Risks Alliance (GRA)AI and Financial Model RiskCredit and pricing models, Explainability and fairness, Generative AI in finance, Human in the loop controls, Model monitoring and drift, Model risk governance, Stress and scenario models, Third party models

Field platforms of the ecosystem where the subject is studied and applied.

Practised in

The three Foundry systems where it is practised

Foundry system pages describe what each system and its rooms are designed to do; the courses teach the methods these systems apply.

Across the ecosystem

Across the ecosystem

The science, platforms and standards this discipline draws on and contributes to across the ecosystem.

Foundations in the STEM Commons

Standards for the field

Pages of the ecosystem, cited so learners can follow the field into research, platforms, standards and national practice.

08অংশ নিন

Take part in Risk Science and Methods

আপনি কে, সেখান থেকে শুরু করুন।

পথ

  • পথ

    জাতীয় প্রতিষ্ঠান

    মন্ত্রণালয়, পরিসংখ্যান দপ্তর ও সংস্থা: একটি জাতীয় ডেস্ক ও শাখা, এবং কার্যকরী দলে একটি আসন। জাতীয় অধ্যায়

  • পথ

    খাত ও পেশা

    Operators, regulators, banks, insurers and professional bodies whose work touches risk science and methods. For organisations

  • পথ

    শিক্ষা

    কোর্স ও অনুশীলন, প্রতিটি সম্পন্ন কোর্সের রেকর্ডসহ। পড়াশোনা বিনামূল্যে। উন্মুক্ত শিক্ষা

  • পথ

    মানদণ্ড

    কার্যকরী দলের সঙ্গে পদ্ধতি ও ডেটা মানদণ্ডের খসড়া তৈরি, পর্যালোচনা ও গ্রহণ করুন। কার্যনির্বাহী দল

  • পথ

    সম্প্রদায় ও গবেষণা

    Universities, civil society and individual experts: projects, data and review. Write to the Academy

উন্মুক্ত আহ্বান

উন্মুক্ত আহ্বানযা খোঁজা হচ্ছেশেষ হয়
কার্যকরী দলের জন্য বিশেষজ্ঞদের প্রতি প্রতিষ্ঠাকালীন আহ্বান CALL-FOUNDING-EXPERTSবিশেষজ্ঞ, সম্পাদক, সভাপতি2026-12-15
বৈশ্বিক মানদণ্ড পরিষদের তিনটি স্বাধীন বিশেষজ্ঞ আসনের জন্য উন্মুক্ত মনোনয়ন CALL-GSC-INDপরিষদ-সদস্য2026-12-01
অন্তর্বর্তী আঞ্চলিক সদস্যদের মনোনয়ন (কেবল অন্তর্বর্তী বিধান সিদ্ধান্ত হলে প্রযোজ্য) CALL-GSC-REGপরিষদ-সদস্য2026-12-15
তরঙ্গ-1 ঝুঁকি মূল রূপরেখার পর্যালোচনা (C1, C2, C3, C4, C8, C10, C11) CALL-REVIEW-WAVE1অবদানকারী2027-01-31
ঝুঁকি বিজ্ঞানের মূল মানদণ্ডের জন্য বিশেষজ্ঞ: অনিশ্চয়তা ও প্রমাণের মাননির্ণয়, স্কেল, কাঠামোবদ্ধ বিশেষজ্ঞ বিচার, সহনীয়তা, চরম ও যৌগিক ঘটনা, বিপর্যয়কর ও অস্তিত্বগত ঝুঁকি, নীতিশাস্ত্র ও ন্যায্যতা, একক ও পরিমাপ (NRS-C17 থেকে NRS-C24) CALL-CORE-SCIENCEবিশেষজ্ঞ, সম্পাদক, সভাপতি2027-01-31
বিপদ অ্যাটলাসের পরিবারগুলোর জন্য স্বাধীন পর্যালোচক, প্রতিটি পরিবারের জন্য অন্তত দুটি অঞ্চল থেকে তিনজন (NHS2-R19), অগ্রাধিকারপ্রাপ্ত পরিবার আগে CALL-ATLAS-REVIEWবিশেষজ্ঞ2027-03-31
অন্তর্বর্তী সীমান্ত নিরাপত্তা পর্যালোচনা প্যানেলের সদস্য ও রেড-টিম পর্যালোচক (জৈবনিরাপত্তা ও জৈবসুরক্ষা, এআই নিরাপত্তা ও সুরক্ষা, গুরুত্বপূর্ণ অবকাঠামোর সুরক্ষা); আবেদন গোপনীয়ভাবে বিবেচিত হয় CALL-FRONTIER-SAFETY-REVIEWবিশেষজ্ঞ2026-11-30
প্রয়োজনীয়তা নিবন্ধ, NHS-2 যন্ত্র উপস্থাপনা, সামঞ্জস্য পরীক্ষা স্যুট ও সামঞ্জস্য মূল্যায়ন প্রকল্পের জন্য অন্টোলজি, মেট্রোলজি, সফটওয়্যার যাচাই ও সামঞ্জস্য মূল্যায়নের বিশেষজ্ঞ CALL-CONFORMANCE-ONTOLOGYবিশেষজ্ঞ, সম্পাদক2027-01-31

সমাপ্তির তারিখ পেরোনো আহ্বান স্বয়ংক্রিয়ভাবে লুকানো হয়। প্রতিটি আহ্বান, প্রস্তাব ও পথ: উন্মুক্ত ক্ষেত্র ও আহ্বান.

সেল ও CERT

  1. সেল গঠন করুন

    পরিপূরক দক্ষতার তিন থেকে সাতজন মানুষকে একত্র করুন: ডেটা, মাঠ অনুশীলন, নীতি ও যোগাযোগ।

  2. নিবন্ধন করুন

    Write to the team with the cell's name, members, area and region. You receive a reference and a contact. Write to the Academy

  3. একটি সনদ বেছে নিন

    একটি কেন্দ্রবিন্দু বেছে নিন: গবেষণা ও প্রমাণ, ডেটা ও মানচিত্র, মানদণ্ড পর্যালোচনা, অথবা সম্প্রদায়ের প্রস্তুতি।

  4. একসঙ্গে শিখুন

    Risk Academy-তে অনুশীলন ও প্রতিটি সম্পন্ন কোর্সের রেকর্ডসহ যৌথ শিক্ষাপথ অনুসরণ করুন। বিপদ ও সহনশীলতা

  5. নেতৃত্ব দিন ও প্রতিবেদন করুন

    Hold regular sessions. Publish what the cell learns through Reports. Bring findings to the working groups. Working groups

  • সম্প্রদায় জরুরি সাড়াদান দল (CERT) হিসেবে বিশেষায়িত হন

    A CERT prepares its neighbourhood for the hazards it faces. It builds a local hazard profile from this site's map. It trains in household and street preparedness. It learns first aid and light search through recognised local providers, and it runs drills. A CERT works alongside the local emergency services and follows their direction. It never replaces them and is not an emergency service.

Index

একটি জাতীয় দল বা স্থানীয় সম্প্রদায় আজ The Global Risks Index (NXS-GRX)-এ যা করতে পারে।

  • NXS-GRX

    Your country's profile

    195টি দেশের উন্মুক্ত ঝুঁকি প্রোফাইল, প্রতিটি মানে উৎস ও লাইসেন্সসহ। সূচক খুলুন

  • NXS-GRX

    সংশোধন বা উত্তর দিন

    আপনার দেশের জন্য একটি মান সংশোধন করুন বা উত্তর দিন। পরিবর্তনটি তার প্রভাবসহ লিপিবদ্ধ হয়। সূচক খুলুন

  • NXS-GRX

    ডেটা দিন

    একটি ডেটাসেট দিন। এটি আপনার স্বীকৃতিসহ লাইসেন্স গেট দিয়ে প্রবেশ করে। সূচক খুলুন

  • NXS-GRX

    উন্মুক্ত ডেটা ব্যবহার করুন

    উন্মুক্ত API-র মাধ্যমে স্তর, প্রোফাইল ও টাইল। সূচক খুলুন

09References

Core references and where to begin

Start with Board of Governors of the Federal Reserve System, then Bank of England Prudential Regulation Authority, then Box.

  1. Board of Governors of the Federal Reserve System, SR 11-7: Guidance on Model Risk Management (2011). federalreserve.gov
  2. Bank of England Prudential Regulation Authority, SS1/23 Model risk management principles for banks (2023). bankofengland.co.uk
  3. National Institute of Standards and Technology, Artificial Intelligence Risk Management Framework (AI RMF 1.0) (2023). DOI
  4. Box, G. E. P., Science and statistics, Journal of the American Statistical Association (1976). DOI
  5. Oreskes, N., Shrader-Frechette, K. and Belitz, K., Verification, validation, and confirmation of numerical models in the earth sciences, Science (1994). DOI
  6. Rudin, C., Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead, Nature Machine Intelligence (2019). DOI
  7. Gama, J., Žliobaitė, I., Bifet, A., Pechenizkiy, M. and Bouchachia, A., A survey on concept drift adaptation, ACM Computing Surveys (2014). DOI
  8. Mitchell, M. et al., Model cards for model reporting, Proceedings of the Conference on Fairness, Accountability, and Transparency (2019). DOI
  9. International Organization for Standardization and IEC, ISO/IEC 42001:2023 Artificial intelligence management system (2023). iso.org

10প্রশ্ন

প্রশ্ন, উত্তর ও মূল পরিভাষা

এখানে অধ্যয়ন কি বিনামূল্যে?

হ্যাঁ। Nexus Consortium-এর নেতৃত্বে পরিচালিত Risk Academy-র প্রতিটি কোর্স ও শিক্ষা উপকরণ অধ্যয়নের জন্য বিনামূল্যে, নিজস্ব গতিতে বা দলবদ্ধভাবে। সাধারণ অংশগ্রহণে কোনো ফি নেই। বেতনভুক্ত কাজ, ইন্টার্নশিপ ও ফেলোশিপ আলাদা ব্যবস্থা।

একটি কোর্স সম্পন্ন করলে আমি কী পাই?

একটি সমাপ্তি নথিতে থাকে কোর্স, তার সংস্করণ, তারিখ ও কী মূল্যায়ন হয়েছে। এটি ডিফল্টভাবে ব্যক্তিগত এবং আপনি তা ভাগ করেন। এটি শাস্ত্র, বিভাগ ও কর্মসূচির অর্জনে জমা হয়। নথিতে কী অধ্যয়ন ও মূল্যায়ন হয়েছে তা বলা থাকে, দক্ষতার প্রত্যয়ন থাকে না; এটি লাইসেন্স, পেশাগত উপাধি বা কর্মসংস্থানের অবস্থা নয়।

তথ্য ও স্থানগুলো কোথা থেকে আসে?

প্রতিটি অনুষদ উন্মুক্ত ডেটা দিয়ে তৈরি একটি বাস্তব স্থানের টুইনে পড়ায়; প্রতিটি স্তরে তার প্রদানকারী, লাইসেন্স ও তারিখ থাকে, এবং একাডেমির প্রতিটি ফলাফলে তার উৎস উল্লেখ থাকে। মানচিত্র ও পরিসংখ্যান শিক্ষাদানের জন্য নির্দেশক, কখনও কোনো স্থানের মূল্যায়ন নয়।

How do I take part beyond the courses in model risk and validation?

তিন থেকে সাতজনের একটি দক্ষতা সেল গঠন করুন, একটি যৌথ পথ অধ্যয়ন করুন, জাতীয় কার্যকরী দলে একটি তত্ত্বাবধানকৃত দায়িত্ব নিন, অথবা আপনার প্রতিষ্ঠান থেকে একটি দল আনুন। প্রতিষ্ঠানের জন্য.

মূল পরিভাষা

মডেল ঝুঁকি
The potential for loss or poor decisions from errors in models or their misuse.
Validation
Independent review that a model is sound and fit for its purpose.
Back-testing
Comparing model predictions with what happened.
Drift
Change in data or relationships that degrades a model over time.

প্রতিবেদনঅবদান রাখুন

এই পাতা থেকে প্রমাণ জমা দিন।

Reports হলো কনসোর্টিয়ামের ত্রৈমাসিক প্রকাশনা। এই বিষয়ে আপনি যা জানেন তা পরবর্তী সংস্করণে প্রমাণ, দৃষ্টান্ত বা সংশোধনী হয়ে উঠতে পারে, আপনার নামে স্বীকৃত হয়ে এবং আপনার নিজস্ব কাজ হলে CC BY 4.0 লাইসেন্সে প্রকাশিত হয়ে।

Risk Academy

কখনো ব্যক্তিগত তথ্য, সংবেদনশীল অবকাঠামোর বিবরণ বা ভাগ করার অধিকার নেই এমন কিছু পাঠাবেন না। কোনো অবদান ব্যবহারের আগে পর্যালোচনা করা হয়; এটি পাঠালে লেখকত্ব বা সম্পাদকীয় অধিকার জন্মায় না, এবং পৃষ্ঠপোষকতা কখনো কোনো সিদ্ধান্ত কেনে না।

মডেল ঝুঁকি ও যাচাই

Prepare with a related course

Open courses in connected disciplines teach the methods this discipline will build on.

কোর্স তালিকা ব্রাউজ করুনBack to Department of Risk Management