index.md

azure-kusto-analytics-lib Repository

1. KqlMagic

   1. Getting Started with KqlMagic

2. Labs

   1. Custom Time Series Forecasting¹ - build and tailor a time series forecasting model from lower level functions. See task readme file
   2. Classification - Build a classifier in Python, score¹ in ADX. See task readme file

3. Machine Learning

   • Functions

     1. classify_sf()¹ - Classify (aka score) samples by a pre-trained model stored in models table
     2. kmeans_sf()¹ - K-Means clustering

   • Queries

     1. Clustering Plugins Tutorial - Walkthrough of the clustering plugins for diagnosis & RCA
     2. Python Plugin Tutorial¹ - Walkthrough and demos of extending Kusto with inline Python

   • Notebooks

     1. Training a classifier¹ - Demo of building and training a classifier for prediction of room occupancy in Jupyter using KqlMagic. Scoring is done using the Python plugin (see the Python tutorial)
     2. Training a classifier in AML and scoring in Kusto¹ - Demo of building and training a classifier for prediction of room occupancy in Azure Machine Learning (AML) service. Training is done on AML cloud compute. Model is exported to Kusto and scoring is done using the Python plugin (see the Python tutorial)
     3. Converting a classifier to ONNX and scoring in Kusto¹ - Demo of building and training a classifier for prediction of room occupancy using Scikit-Learn. The trained model is converted to ONNX format and exported to Kusto. Scoring is done in Kusto using the Python plugin (see the Python tutorial)
     4. Training a classifier in Synapse Spark pool, scoring in Data Explorer (Kusto) pool¹ - Demo of building and training a classifier for prediction of room occupancy using Spark ML on Synapse Spark pool. The trained model is converted to ONNX format and exported to Data Explorer (Kusto) pool, on the same Synapse workspace. Scoring is done in Data Explorer using the Python plugin (see the Python tutorial)

4. Time Series Analysis

   • Functions

     1. blackman_filter_sf() - Create a Blackman window low pass filter of specific width
     2. series_fit_poly_sf()¹ - Fit a polynomial of a specified degree to a series
     3. series_fit_sine_sf()¹ - Fit a sine wave to a series
     4. series_moving_avg_sf() - Moving average of specific width
     5. series_partial_sf() - Test for series with empty bins
     6. series_rolling_sf()¹ - Rolling window functions on a series
     7. series_segment_sf() - Sequental numbering of non zero segments of a boolean series
     8. series_summarize_sf()¹ - Aggregation functions on a series

   • Queries

     1. Time Series Analysis Tutorial - Walkthrough of typical series functions from each category

5. Utils

   • Functions
     1. get_df_schema_sf()¹ - Returns the schema of df (the DataFrame of the input table)
     2. get_modules_version_sf()¹ - Returns version information for the Python engine and the specified packages

6. Large Language Models (LLM)

   • Notebooks

     1. Vector Semantic Search with Open AI - Demo of vector semantic search with OpenAI and Kusto.

Notes:

¹ Implementation using inline Python plugin