Scribd
Upload
81 views7 pages

Snowflake MLPF

Snowflake MLPF (Machine Learning Powered Functions) provide easy to use SQL functions powered by machine learning to help analysts make better, faster decisions. The functions include forecasting, anomaly detection, and contribution explorer. Anomaly detection uses machine learning to identify outliers in time series data and trigger alerts. It allows predicting anomalies for single or multiple time series, including exogenous variables, labeling anomalies, and adjusting the prediction interval. The underlying algorithm is gradient boosting machine. The function provides options like separate monitoring of series and using labeled anomalies to improve accuracy.

Uploaded by

jahnavi208
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
81 views7 pages

Snowflake MLPF

Snowflake MLPF (Machine Learning Powered Functions) provide easy to use SQL functions powered by machine learning to help analysts make better, faster decisions. The functions include forecasting, anomaly detection, and contribution explorer. Anomaly detection uses machine learning to identify outliers in time series data and trigger alerts. It allows predicting anomalies for single or multiple time series, including exogenous variables, labeling anomalies, and adjusting the prediction interval. The underlying algorithm is gradient boosting machine. The function provides options like separate monitoring of series and using labeled anomalies to improve accuracy.

Uploaded by

jahnavi208
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 7

Snowflake MLPF

(Machine Learning Powered


Functions)
What are MLPF
functions?
•Easy to use SQL functions
powered by ML.

•Abstract away the complexity of


ML frameworks to empower
analysts to make better decisions,
faster

•Forecasting, Anomaly Detection,


Contribution Explorer in Public
Preview at Summit
Benefits

ML accessible via SQL No complex Infrastructure Quick insights delivery


Leverage SQL functions that Scale from one or millions of ML- Deliver results through familiar
abstract the complexity of ML powered insights with the elasticity analytics and BI tools integrated
frameworks and algorithms to and near-zero operations of the with snowflake’s consistent data
effortlessly enhance decision speed snowflake processing engine governance across model inputs
and quality and outputs
What can MLPF do?

Forecasting Anomaly Detection Contribution Explorer


Use ML to build more reliable time Use machine learning to identify Use machine learning to quickly
series forecasts with the automated outliers in time series and trigger identify top drivers contributing to
handling of seasonality, missing alerts or help find unlikely-to- the change of a given metrics across
values and more happen events that should either be user-defined time intervals
excluded form the analysis or should
be focused on
What it Does – Anomaly Detection

• Use ML to identify outliers in time series and trigger alerts or help find unlikely-to-happen
events that should be either be excluded from the analysis or should be focused on
• Key functionalities:
 Predicting anomalies for a single time series or multi-series
 Including exogenous variables and exploring their relative importance
 Labeling anomalies and generating new predications with identified anomalies
( supervised predictions)
 Setting how rare you want your identified anomalies to be (1% of forecasted possibilities)
 The underlying algorithm is GBM ( Gradient boosting machine)
• This function provides options for building on a simple single-
series detection:

• You can use column in your data table to identify series that
should be monitored separately for anomalies ( e.g

How it works storeA_jacket vs storeB_jacket)

– Anomaly • You can provide the model with hand-labeled anomalies in


Detection your training data - to ensure future anomalies are identified as
accurately as possible

• You can determine the size of your predication interval. A


smaller prediction interval corresponds to a stricter definition
of “normal” - and increases the number of anomalies you’ll
identify.
Link to the Snowflake documentation:
https://docs.snowflake.com/en
/user-guide/ml-powered-anomaly-detection#label-analysis-a
nomaly-detection-examples

You might also like