Introducing a new data analysis package
Recently, I had a meeting with a great friend who has a successful company in AI and data analysis field. He introduced an online platform related to data analysis, and I found it very interesting.
In this article, I will talk about this platform and its different aspects. Note that this is my first time working with this tool. Let’s go and find the features together.
The name is Shapelets, and you can Google it. Its mission is to facilitate data analysis for the data scientist with online tools such as data entry, storage, analysis, and data visualization.
You can import the data to the platform from your database or other databases as well.
One interesting point is this platform also offers different datasets in its storage and you can move your data to its storage as well.
Shapelets provides different services including memory-centric APIs, storage, streaming, and execution servers. You can use their server to run pre-defined machine learning models in GPU’s parallel processing in a short time.
It also provides the EDA tools such as feature engineering and studying the relationship between the variables. It also supplies time series tools in its streaming part
In the end, the user has access to the final reports including the plots, graphs, analysis reports, BI tools, and Microsoft office integrations.
One thing that I liked as a developer is that I can develop my algorithm and merge it with their pre-defined algorithms. I should thank my friend who told me about this platform.
You can find some use cases in the solution tab for different categories including, IoT, Energy, or aerospace.
You can install the package through the pip
pip install shapelets-compute
One of the requirements is ArrayFire library which is designed to work on GPU processing.
install run time dependencies
shapelets install <<backend>>
Shapelet has its own array wrapper type built over ArrayFire. please refer here to create an array.
After installing the shapelets, you can import the library by calling
If it returns the following error, you have to upgrade your ArrayFire package.
No module named ‘shapelets.compute‘
I will try to implement some models through this platform in the coming days, so don’t forget to check my GitHub.
In future reviews, I will include some case studies as well.