Intel® oneAPI Data Analytics Library for Python* Examples

ID 标签 659727
已更新 3/4/2020
版本 Latest
公共

author-image

作者

Introduction

Intel® oneAPI Data Analytics Library (oneDAL) is a highly optimized data analytics and machine learning (ML) library.  It provides C++, Java*, and Python* interfaces, as well as newly introduced Data Parallel C++ (DPC++)  interface.  

The library with Python interface is also called DAAL4py. It is available along with Intel® Distribution for Python* (IDP), which is part of Intel® oneAPI Base Toolkit and Intel® AI Analytics Toolkit, as well as available from open-source repository: DAAL4py GitHub* repository.

Intel® oneDAL DAAL4py Public Samples

In order to help users to understand the Intel® AI Analytics Toolkit product,  we have officially published oneAPI DAAL4py samples at AIkit code-sample. For example:

  • Intel oneAPI Data Analytics Library Hello World (daal4py)
  • Intel oneAPI Data Analytics Library for Distributed Linear Regression Training and Prediction (daal4py)
  • Intel oneAPI Data Analytics Library for Distributed K-Means Training and Prediction (daal4py) 

View All Code Samples

Intel® DAAL4py Self-Owned Examples

In order to help users to understand the Intel DAAL4py product, we also provide a suit of examples to demonstrate the product functionality. For example:  

  • Decision Forest for Classification and Regression
  • Gradient Boosted Trees (GBT) for Classification and Regression
  • K-Means Clustering
  • K-Nearest Neighbor (KNN)
  • Logistic Regression 
  • Multinomial Naïve Bayes Classifier
  • Support Vector Machines (SVM) with Linear and Radial Basis Function (RBF) Kernels
  • Multiclass Classification Using a One-Against-One Strategy
  • Principal Components Analysis (PCA)
  • And more

These samples are available from both the Intel oneAPI Base Toolkit and Intel® AI Analytics Toolkit, as well as the DAAL4py GitHub* repository. 

Intel DAAL4py-related Samples and Products

Thus, we have three DAAL4py-related samples and products at hand. See the table:

Product

Product Samples

Resources /Location

Intel® AI Analytics Toolkit(beta)

AIKit Code Samples https://github.com/intel/AiKit-code-samples

Intel oneAPI Base Toolkit(beta)

DAAL4py Self-owned Examples <oneAPI INSTALLATION PATH> /Intelpyton/latest/pkgs/daal4py-xxx/info/test/examples

DAAL4py GitHub* Repository  

DAAL4py Self-owned Examples in GitHub* 

https://github.com/IntelPython/daal4py/tree/master/examples

In order to avoid confusion, we encourage you to use the matching product samples for a product.

Troubleshoot

If you try the samples in a different product other than the intended one, errors may arise like below, 

Unexpected Error: TypeError: __cinit__() got an unexpected keyword argument 'votingMethod'

To run the DAAL4py Self-owned Example in GitHub* with Intel® oneAPI Base Toolkit product

#Set oneAPI environment: 
source /opt/intel/inteloneapi/setvars.sh
#But using the DAAL4py Example from the DAAL4py public GitHub* repository:
$git clone https://github.com/IntelPython/daal4py.git
$cd daal4py/examples
$ python decision_forest_classification_batch.py
Error Message
Traceback (most recent call last):
  File "decision_forest_classification_batch.py", line 76, in <module>
    (train_result, predict_result, plabels) = main()
  File "decision_forest_classification_batch.py", line 62, in main
    resultsToEvaluate="computeClassLabels|computeClassProbabilities", votingMethod="unweighted")
  File "build/daal4py_cy.pyx", line 12663, in _daal4py.decision_forest_classification_prediction.__cinit__
TypeError: __cinit__() got an unexpected keyword argument 'votingMethod'

Root Cause

We keep developing new algorithms and committing the latest changes to DAAL4py GitHub* repository. Not all of the latest changed are applicable to other products. Hence, it is wrong to check the DAAL4py GitHub* examples with the oneAPI product.   

Resolution

Use matched sample with its intended product. For example, to use the self-owned DAAL4py examples with the Intel oneAPI Base Toolkit(beta):

$ source /opt/intel/inteloneapi/setvars.sh
$ cd /opt/intel/inteloneapi/intelpython/latest/pkgs/daal4py-2021.1b4-py37ha68da19_4/info/test/examples/
$ python decision_forest_classification_batch.py
Variable importance results:
 [[78.0208408  79.09469066  0.27575422]]
......
OOB error:
 [[0.03299533]]
PASSED