Thanks Maheshakya and Manorama for carrying out these tests. Your feedback
is welcome.

On Sun, Sep 13, 2015 at 2:29 PM, Maheshakya Wijewardena <[email protected]
> wrote:

> Hi,
>
> We have measured the average times taken by the ML algorithms to build
> models and predict for different sizes of datasets. Following dataset sizes
> were used.
>
>
>    - 100
>    - 1000
>    - 10000
>    - 100000
>    - 1000000
>
> These different size datasets were created by randomly re-sampling the
> data points from the following publicly available datasets:
>
>    1. Poker hand dataset[1]  - For multi class classification, numerical
>    prediction and clustering.
>    2. Higgs Atlas dataset sample dataset [2] - For binary classification
>
> Number of features is kept fixed at 10.
>
> For model building, average build times were calculated across 3
> iterations for each algorithm for each dataset size. Prediction times are
> the cumulated time took to predict for all the rows in the datasets.
>
> Graphs depicting the performance measurements are attached herewith. In
> the graphs, X axes are indented in log scale.
>
>
> Best regards.
>
> [1] http://archive.ics.uci.edu/ml/datasets/Poker+Hand
>
> [2] http://opendata.cern.ch/record/328?ln=en
>
> --
> Pruthuvi Maheshakya Wijewardena
> Software Engineer
> WSO2 : http://wso2.com/
> Email: [email protected]
> Mobile: +94711228855
>
>
>
> _______________________________________________
> Architecture mailing list
> [email protected]
> https://mail.wso2.org/cgi-bin/mailman/listinfo/architecture
>
>


-- 

Thanks & regards,
Nirmal

Team Lead - WSO2 Machine Learner
Associate Technical Lead - Data Technologies Team, WSO2 Inc.
Mobile: +94715779733
Blog: http://nirmalfdo.blogspot.com/
_______________________________________________
Architecture mailing list
[email protected]
https://mail.wso2.org/cgi-bin/mailman/listinfo/architecture

Reply via email to