# NAB

**URL:** https://discourse.numenta.org/c/engineering/nab/36.md

[Latest](https://discourse.numenta.org/latest.md) · [Categories](https://discourse.numenta.org/categories.md) · [Tags](https://discourse.numenta.org/tags.md)

---

## [About the NAB category](https://discourse.numenta.org/t/about-the-nab-category/438)

<div class="topic-metadata">

**Author:** [@rhyolight](https://discourse.numenta.org/u/rhyolight)\
**Replies:** 0

</div>

The Numenta Anomaly Benchmark is a novel benchmark for evaluating algorithms for anomaly detection in streaming, real-time applications.

---

## [Htm.core vs NuPIC on NAB](https://discourse.numenta.org/t/htm-core-vs-nupic-on-nab/10207)

<div class="topic-metadata">

**Author:** [@sheiser1](https://discourse.numenta.org/u/sheiser1)\
**Replies:** 9\
**Last updated:** [November 16, 2022, 5:10am UTC](https://discourse.numenta.org/t/htm-core-vs-nupic-on-nab/10207 "2022-11-16T05:10:02Z")

</div>

Hey all, So I’m trying to debug the difference in NAB score between NuPIC (as used by the NumentaTM detector) and htm.core. In my implementation I have aligned all the model hyperparams (SP, TM & Encoders), and also us…

---

## [ARTime leading the Numenta anomally benchmark](https://discourse.numenta.org/t/artime-leading-the-numenta-anomally-benchmark/9511)

<div class="topic-metadata">

**Author:** [@markNZed](https://discourse.numenta.org/u/markNZed)\
**Replies:** 0\
**Last updated:** [February 25, 2022, 10:37am UTC](https://discourse.numenta.org/t/artime-leading-the-numenta-anomally-benchmark/9511 "2022-02-25T10:37:52Z")

</div>

Thanks to @subutai and @lscheinkman the ARTime results have been released in the official NAB scoreboard GitHub - numenta/NAB: The Numenta Anomaly Benchmark where ARTime is leading at the moment. It says a lot about Nume…

---

## [New NAB Detector ARTime](https://discourse.numenta.org/t/new-nab-detector-artime/9507)

<div class="topic-metadata">

**Author:** [@markNZed](https://discourse.numenta.org/u/markNZed)\
**Replies:** 3\
**Last updated:** [February 23, 2022, 9:36pm UTC](https://discourse.numenta.org/t/new-nab-detector-artime/9507 "2022-02-23T21:36:42Z")

</div>

The ARTime detector is available in a PR to NAB. The results are available in the ARTimeNAB repo. To summarize: Detector Standard Profile Reward Low FP Reward Low FN Perfect 100.0 100.0 100.0 ARTime 74.9 65.1 8…

---

## [Impact on NAB Scoreboard of correcting label window](https://discourse.numenta.org/t/impact-on-nab-scoreboard-of-correcting-label-window/9147)

<div class="topic-metadata">

**Author:** [@markNZed](https://discourse.numenta.org/u/markNZed)\
**Replies:** 0\
**Last updated:** [November 11, 2021, 7:03pm UTC](https://discourse.numenta.org/t/impact-on-nab-scoreboard-of-correcting-label-window/9147 "2021-11-11T19:03:40Z")

</div>

I created an issue about an erroneous window label in Erroneous window · Issue #385 · numenta/NAB · GitHub and I’ve submitted a PR to fix that at Corrected window for realAWSCloudwatch/iio\_us-east-1\_i-a2eb1cd9\_Netwo… by …

---

## [Overconstraining Threshold?](https://discourse.numenta.org/t/overconstraining-threshold/9071)

<div class="topic-metadata">

**Author:** [@markNZed](https://discourse.numenta.org/u/markNZed)\
**Replies:** 0\
**Last updated:** [October 15, 2021, 6:02am UTC](https://discourse.numenta.org/t/overconstraining-threshold/9071 "2021-10-15T06:02:04Z")

</div>

From the NAB whitepaper “We constrain the detectors to use a single detection threshold for the entire dataset.” I am not sure this is a reasonable approach for a community benchmark because some algorithms may generate …

---

## [Label Column in the detection phase output](https://discourse.numenta.org/t/label-column-in-the-detection-phase-output/5783)

<div class="topic-metadata">

**Author:** [@Teerarat\_Siwapathomc](https://discourse.numenta.org/u/Teerarat_Siwapathomc)\
**Replies:** 3\
**Last updated:** [October 14, 2021, 2:21pm UTC](https://discourse.numenta.org/t/label-column-in-the-detection-phase-output/5783 "2021-10-14T14:21:57Z")

</div>

I used the data from NAB, for example “numenta\_nyc\_taxi.csv” to run my own detector outside. My detector didn’t generate the labels field. Based on my understanding, the labels field wont change if we use the same data …

---

## [Threshold calculation and windows](https://discourse.numenta.org/t/threshold-calculation-and-windows/9014)

<div class="topic-metadata">

**Author:** [@markNZed](https://discourse.numenta.org/u/markNZed)\
**Replies:** 0\
**Last updated:** [October 2, 2021, 8:12am UTC](https://discourse.numenta.org/t/threshold-calculation-and-windows/9014 "2021-10-02T08:12:29Z")

</div>

In a given anomaly window only the first anomaly indication is counted. Is the threshold calculation taking this into acount ? I could not see that in the code after a quick look. It might impact the scores if the algori…

---

## [Timesteps in NAB](https://discourse.numenta.org/t/timesteps-in-nab/8946)

<div class="topic-metadata">

**Author:** [@markNZed](https://discourse.numenta.org/u/markNZed)\
**Replies:** 3\
**Last updated:** [September 17, 2021, 6:47am UTC](https://discourse.numenta.org/t/timesteps-in-nab/8946 "2021-09-17T06:47:28Z")

</div>

Is it intentional that some time series in NAB do not have regular time steps e.g. realAWSCloudwatch/ec2\_cpu\_utilization\_825cc2.csv lines 39, 1116?

---

## [Global knowledge provided to the detector in NAB](https://discourse.numenta.org/t/global-knowledge-provided-to-the-detector-in-nab/8886)

<div class="topic-metadata">

**Author:** [@markNZed](https://discourse.numenta.org/u/markNZed)\
**Replies:** 1\
**Last updated:** [August 20, 2021, 12:16pm UTC](https://discourse.numenta.org/t/global-knowledge-provided-to-the-detector-in-nab/8886 "2021-08-20T12:16:59Z")

</div>

As an example of global knowledge, the detector can use information about the range of values in each data set. Is the detector also told how many initial samples are gauranteed not to contain anomalies? I saw that it is…

---

## [NAB Application Profiles](https://discourse.numenta.org/t/nab-application-profiles/8474)

<div class="topic-metadata">

**Author:** [@pmoriano](https://discourse.numenta.org/u/pmoriano)\
**Replies:** 0\
**Last updated:** [April 22, 2021, 12:48pm UTC](https://discourse.numenta.org/t/nab-application-profiles/8474 "2021-04-22T12:48:24Z")

</div>

Hello, I wonder where I can find the values used in each application profile A in the NAB paper. In particular, what are the values used in the paper for A\_{TP}, A\_{FP}, A\_{FN}, A\_{TN} under the standard, reward low FP,…

---

## [NAB Normalized Scores](https://discourse.numenta.org/t/nab-normalized-scores/8509)

<div class="topic-metadata">

**Author:** [@pmoriano](https://discourse.numenta.org/u/pmoriano)\
**Replies:** 0\
**Last updated:** [April 28, 2021, 6:50pm UTC](https://discourse.numenta.org/t/nab-normalized-scores/8509 "2021-04-28T18:50:16Z")

</div>

Hello, I am having trouble to interpret the normalized NAB score (Eq. (4) in the conference paper). The result is supposed to be in the range \[0, 1\], but this depends on how many false positives you get, i.e., if you ge…

---

## [Why does the spatial pooler for the NAB HTM detector use wraparound when the input is not circular?](https://discourse.numenta.org/t/why-does-the-spatial-pooler-for-the-nab-htm-detector-use-wraparound-when-the-input-is-not-circular/7911)

<div class="topic-metadata">

**Author:** [@nothisispatrick](https://discourse.numenta.org/u/nothisispatrick)\
**Replies:** 1\
**Last updated:** [January 5, 2021, 8:43am UTC](https://discourse.numenta.org/t/why-does-the-spatial-pooler-for-the-nab-htm-detector-use-wraparound-when-the-input-is-not-circular/7911 "2021-01-05T08:43:14Z")

</div>

The parameters for the NAB HTM detector are specified here. Notice that wrapAround is not specified, which means it defaults to True (src/nupic/alogrithms/spatial\_pooler.py#L251 – sorry I’m new so I can only post 2 links…

---

## [Evaluating the performance of the real-time anomaly detection included in the Elastic Stack](https://discourse.numenta.org/t/evaluating-the-performance-of-the-real-time-anomaly-detection-included-in-the-elastic-stack/7139)

<div class="topic-metadata">

**Author:** [@Dima](https://discourse.numenta.org/u/Dima)\
**Replies:** 9\
**Last updated:** [March 16, 2020, 7:52pm UTC](https://discourse.numenta.org/t/evaluating-the-performance-of-the-real-time-anomaly-detection-included-in-the-elastic-stack/7139 "2020-03-16T19:52:45Z")

</div>

I am about to do my MSc thesis and I have in mind to do a comparative study between the Anomaly Detection algorithm included in the Elastic Stack and 1-2 other Frameworks/Implementations. I was wondering whether it is p…

---

## [Difference between combined\_windows.json and combined\_labels.json](https://discourse.numenta.org/t/difference-between-combined-windows-json-and-combined-labels-json/7067)

<div class="topic-metadata">

**Author:** [@iliabautista](https://discourse.numenta.org/u/iliabautista)\
**Replies:** 1\
**Last updated:** [January 20, 2020, 7:23pm UTC](https://discourse.numenta.org/t/difference-between-combined-windows-json-and-combined-labels-json/7067 "2020-01-20T19:23:40Z")

</div>

In https://github.com/numenta/NAB/tree/master/labels, there are two types of labels: combined\_windows.json and combined\_labels.json. Is there a difference between the two? Thanks in advance.

---

## [NAB v1.1 Release](https://discourse.numenta.org/t/nab-v1-1-release/6904)

<div class="topic-metadata">

**Author:** [@rhyolight](https://discourse.numenta.org/u/rhyolight)\
**Replies:** 0\
**Last updated:** [December 11, 2019, 6:14pm UTC](https://discourse.numenta.org/t/nab-v1-1-release/6904 "2019-12-11T18:14:08Z")

</div>

We have released NAB v1.1, which updates NAB to Python 3.

---

## [Comparison between HTM and other anomaly detection algorithms](https://discourse.numenta.org/t/comparison-between-htm-and-other-anomaly-detection-algorithms/6855)

<div class="topic-metadata">

**Author:** [@artifex](https://discourse.numenta.org/u/artifex)\
**Replies:** 3\
**Last updated:** [November 29, 2019, 5:12am UTC](https://discourse.numenta.org/t/comparison-between-htm-and-other-anomaly-detection-algorithms/6855 "2019-11-29T05:12:01Z")

</div>

Hi. I have used the HTM for several private, off-work projects and I find it really great. Today I had a meeting with my CEO and I suggested that we’ll use the HTM algorithm for some project we’ve been working on here a…

---

## [Why did i get double TP ( true positive ) than the NAB source on github](https://discourse.numenta.org/t/why-did-i-get-double-tp-true-positive-than-the-nab-source-on-github/6844)

<div class="topic-metadata">

**Author:** [@life\_happy](https://discourse.numenta.org/u/life_happy)\
**Replies:** 3\
**Last updated:** [November 27, 2019, 10:48pm UTC](https://discourse.numenta.org/t/why-did-i-get-double-tp-true-positive-than-the-nab-source-on-github/6844 "2019-11-27T22:48:56Z")

</div>

I run HTM NUPIC ( python 2.7) with API algorithm. i run with data https://github.com/numenta/NAB/blob/master/data/realKnownCause/machine\_temperature\_system\_failure.csv and have the result ( i capture from my file) I…

---

## [Noobie Question: How to use NuPIC for a NAB dataset?](https://discourse.numenta.org/t/noobie-question-how-to-use-nupic-for-a-nab-dataset/5005)

<div class="topic-metadata">

**Author:** [@flarelink](https://discourse.numenta.org/u/flarelink)\
**Replies:** 9\
**Last updated:** [November 23, 2019, 8:53am UTC](https://discourse.numenta.org/t/noobie-question-how-to-use-nupic-for-a-nab-dataset/5005 "2019-11-23T08:53:34Z")

</div>

Hello everyone, I took some time off from working on HTM due to some other work but have come back to take a look at it again for streaming anomaly detection tasks. I’ve been reading some HTM papers and I have also come…

---

## [NAB scoring returns minus score for each true negative](https://discourse.numenta.org/t/nab-scoring-returns-minus-score-for-each-true-negative/6819)

<div class="topic-metadata">

**Author:** [@michellea](https://discourse.numenta.org/u/michellea)\
**Replies:** 3\
**Last updated:** [November 18, 2019, 7:20pm UTC](https://discourse.numenta.org/t/nab-scoring-returns-minus-score-for-each-true-negative/6819 "2019-11-18T19:20:16Z")

</div>

Hi there, I am implementing NAB with as final goal to start testing with HTM. Right now, I am just trying to implement the checked out version of april this year of the NAB github. I started running it with the random d…

---

## [Same parameter get different result when using BacktrackingTMCPP](https://discourse.numenta.org/t/same-parameter-get-different-result-when-using-backtrackingtmcpp/6579)

<div class="topic-metadata">

**Author:** [@bcc](https://discourse.numenta.org/u/bcc)\
**Replies:** 4\
**Last updated:** [September 16, 2019, 9:29pm UTC](https://discourse.numenta.org/t/same-parameter-get-different-result-when-using-backtrackingtmcpp/6579 "2019-09-16T21:29:24Z")

</div>

Hi,I am trying to use algorithm api to detect anomaly on nyc\_taxi.csv. But my result is poor.So i try to repeat the NAB result. I use the same parameter. The sp’s output is same to NAB’s. However BacktrackingTMCPP is not…

---

## [Data pre processing with different models in NAB](https://discourse.numenta.org/t/data-pre-processing-with-different-models-in-nab/6421)

<div class="topic-metadata">

**Author:** [@yannc](https://discourse.numenta.org/u/yannc)\
**Replies:** 2\
**Last updated:** [August 6, 2019, 9:31am UTC](https://discourse.numenta.org/t/data-pre-processing-with-different-models-in-nab/6421 "2019-08-06T09:31:14Z")

</div>

Hi there, I am working on comparisons between LSTM using different data reduction techniques and models results given by the NAB. As my focus is computing cost saving, I am interested to know more in detail process tha…

---

## [Naive predictor: how compute anomaly score from 2 scalars? Include error metrics into NAB? Is anomaly score the optimal metric?](https://discourse.numenta.org/t/naive-predictor-how-compute-anomaly-score-from-2-scalars-include-error-metrics-into-nab-is-anomaly-score-the-optimal-metric/6277)

<div class="topic-metadata">

**Author:** [@breznak](https://discourse.numenta.org/u/breznak)\
**Replies:** 2\
**Last updated:** [July 10, 2019, 3:26pm UTC](https://discourse.numenta.org/t/naive-predictor-how-compute-anomaly-score-from-2-scalars-include-error-metrics-into-nab-is-anomaly-score-the-optimal-metric/6277 "2019-07-10T15:26:01Z")

</div>

Morning good fellas, I have begun fiddling with NAB and have 2 questions for you. Technical: “How do I compute anomaly score for time-series from only 2 scalars: current, predicted?” I have implemented a simple, yet ha…

---

## [Model Param for NAB](https://discourse.numenta.org/t/model-param-for-nab/5845)

<div class="topic-metadata">

**Author:** [@Teerarat\_Siwapathomc](https://discourse.numenta.org/u/Teerarat_Siwapathomc)\
**Replies:** 2\
**Last updated:** [April 22, 2019, 1:51am UTC](https://discourse.numenta.org/t/model-param-for-nab/5845 "2019-04-22T01:51:29Z")

</div>

When I run the nupic, I optimize the model param by using swarm optimization but I cannot see the numenta\_detector in NAB use it. Where the numenta get the model param? I saw the this section; def \_setupEncoderParams…

---

## [NAB: codependent metrics from a single system?](https://discourse.numenta.org/t/nab-codependent-metrics-from-a-single-system/5607)

<div class="topic-metadata">

**Author:** [@Oblynx](https://discourse.numenta.org/u/Oblynx)\
**Replies:** 6\
**Last updated:** [March 17, 2019, 8:10am UTC](https://discourse.numenta.org/t/nab-codependent-metrics-from-a-single-system/5607 "2019-03-17T08:10:44Z")

</div>

Does any of the NAB datasets provide different metrics from the same physical system, that would therefore provide different views into the same behavior? For example, do the different realAWSCloudwatch metrics come from…

---

## [Where I can find swarmed parameters for each data sets in NAB?](https://discourse.numenta.org/t/where-i-can-find-swarmed-parameters-for-each-data-sets-in-nab/5534)

<div class="topic-metadata">

**Author:** [@Teerarat\_Siwapathomc](https://discourse.numenta.org/u/Teerarat_Siwapathomc)\
**Replies:** 2\
**Last updated:** [February 28, 2019, 4:39pm UTC](https://discourse.numenta.org/t/where-i-can-find-swarmed-parameters-for-each-data-sets-in-nab/5534 "2019-02-28T16:39:34Z")

</div>

Where I can find the parameters have been used in HTM model for each data sets in NAB repos?

---

## [NAB: faster optimization and scoring](https://discourse.numenta.org/t/nab-faster-optimization-and-scoring/5101)

<div class="topic-metadata">

**Author:** [@Balladeer](https://discourse.numenta.org/u/Balladeer)\
**Replies:** 10\
**Last updated:** [January 10, 2019, 5:18pm UTC](https://discourse.numenta.org/t/nab-faster-optimization-and-scoring/5101 "2019-01-10T17:18:40Z")

</div>

Hey folks! I’m still plugging away at improvements to my HTM implementation. In the process, I’ve been taking a look at NAB! I’ve had a lot of fun digging into the code, particularly the scorer. Short version I think …

---

## [Anomaly Detection for Multivariate TimeSeries Data](https://discourse.numenta.org/t/anomaly-detection-for-multivariate-timeseries-data/5107)

<div class="topic-metadata">

**Author:** [@Sushant](https://discourse.numenta.org/u/Sushant)\
**Replies:** 2\
**Last updated:** [December 31, 2018, 3:26pm UTC](https://discourse.numenta.org/t/anomaly-detection-for-multivariate-timeseries-data/5107 "2018-12-31T15:26:43Z")

</div>

Hi Everyone, I am new to HTM and has gone through basics of HTM. I want to use HTM for Anomaly Detection for Multivariate TimeSeries Data. I got Hot Gym sample code for univariate data. It will be very useful to under…

---

## [HTM Implementation](https://discourse.numenta.org/t/htm-implementation/4910)

<div class="topic-metadata">

**Author:** [@kourosh](https://discourse.numenta.org/u/kourosh)\
**Replies:** 3\
**Last updated:** [November 9, 2018, 10:55am UTC](https://discourse.numenta.org/t/htm-implementation/4910 "2018-11-09T10:55:04Z")

</div>

Hello I am new in HTM. I have tried to use NAB as a part of implementation in my thesis but got confused as I saw python codes. Is there any instruction which shows how to implement NAB??

---

## [Swarm.py with a dataset from NAB](https://discourse.numenta.org/t/swarm-py-with-a-dataset-from-nab/4790)

<div class="topic-metadata">

**Author:** [@andrewfungli](https://discourse.numenta.org/u/andrewfungli)\
**Replies:** 4\
**Last updated:** [October 23, 2018, 5:18pm UTC](https://discourse.numenta.org/t/swarm-py-with-a-dataset-from-nab/4790 "2018-10-23T17:18:03Z")

</div>

Hello, I am trying to implement HTM on another dataset (more specifically: Twitter\_volume\_FB.csv) from the NAB dataset by following the Youtube video example (hot gym prediction model). I have nupic set up on Ubuntu 14,…

[Next page](https://discourse.numenta.org/c/engineering/nab/36.md?page=1)
