Streamlining Data Science with MLJAR Studio
A Guide for Accelerated Machine Learning Workflows
Data science can be significantly challenging, especially for beginners in the field. Statistical analysis, data visualization and machine learning are complicated disciplines, requiring attention to detail and significant expertise. Data professionals should not become frustrated though, as modern tools can help them simplify complicated tasks, thus saving valuable time. For example, MLJAR is a leading data science company that has developed powerful software to assist data scientists. In this article, we are utilizing various MLJAR tools to create a streamlined data science workflow. More specifically, we are going to explore the functionality of AutoML and MLJAR Studio . Let’s begin working on the project!
MLJAR AutoML is an open source Python library aiming to automate machine learning experiments. AutoML provides a complete workflow to preprocess the data, train machine learning models and perform hyper-parameter tuning to find the best one. The library supports advanced automated machine learning capabilities, thus enabling optimized exploratory data analysis and ML model training. Furthermore, AutoML has been cited in more than 100 research papers, hence being established as a scientific tool. You can easily install MLJAR AutoML by running the following command on your terminal:
pip install mljar-supervisedIf you prefer the conda package manager, enter this command instead :
conda install -c conda-forge mljar-supervisedAutoML can be easily used in any Python environment, but we are going to work with MLJAR Studio in this tutorial. Keep in mind that Studio includes every necessary library, so you don’t need to install AutoML separately!
MLJAR Studio is an amazing data science toolkit that accelerates your data science workflows by utilizing a powerful AI assistant. MLJAR provides free access to their customized AI model, specifically trained for data science tasks. Alternatively, you can also use OpenAI models or local AI with Ollama. Here are the main features of MLJAR Studio:
AI Data Analyst: this is an AI-powered assistant utilized for notebook-based data analysis. You start by asking questions in natural language, then apply analysis on local datasets, and get Python code that you can inspect and execute!
AI Code Assistant: this is an AI agent that helps you write code in data science projects. Whether you need help with data cleaning, visualization, or machine learning, the AI assistant provides optimized Python code for every task. The code blocks are easy to review, edit, and integrate right inside your Jupyter notebook.
AutoLab Experiments: AutoLab Experiments let you run machine learning experiments in a structured workflow. You begin by defining a task, set the constraints, and let AI agents explore multiple solutions while keeping everything transparent and reproducible.
In the rest of this tutorial, we are going to explore practical use cases and workflows for each of the MLJAR Studio features. First of all, you need to download and install the software by visiting this link. MLJAR is compatible with all major operating systems, including Microsoft Windows, Apple macOS and Linux. After installing the software, you have to install the Python environment and start using the free plan, or consider upgrading to pro / business. You are then ready to begin working on your data science projects!
AI Data Analyst
As mentioned previously, the AI Data Analyst helps you analyze the chosen dataset of your preference, by using natural language prompts. You can begin using this feature by clicking Create a New Analysis in the main menu, and then selecting AI Data Analyst. We then enter a prompt, asking the AI Data Analyst to load, analyze and visualize the Iris dataset. Iris is one of the most popular datasets used in statistics and machine learning, containing 3 classes that refer to Iris plant species, i.e. Setosa, Versicolour and Virginica.
As demonstrated in the video above, AI Data Analyst successfully loads the Iris dataset, then displays basic information about it and creates the requested figure. Every Jupyter notebook cell is explained in detail, so we comprehend the included Python code. Finally, the data analysis insights are presented, thus helping us understand our dataset, as well as the AI model output in general. By default, the Python code and detailed notebook cell explanations are hidden, hence letting users focus on the data insights exclusively, if they prefer to choose that option. Therefore, AI Data Analyst is an excellent tool for business users who require insights, without necessarily being familiar with the technical aspect of Python coding.
AI Code Assistant
After creating a project with the AI Data Analyst, we can simply click Open in AI-Assisted Notebook to enable this mode. Alternatively, we can open the main menu of MLJAR Studio and select AI-Assisted Notebook, to create a new Jupyter file (.ipynb). In contrast to AI Data Analyst which only supports text prompts, this mode is a fully functional Jupyter notebook environment. Therefore, you can add and execute cells with Python code, or include Markdown text cells. For additional information about using Jupyter notebooks, please read the JupyterLab documentation. We are now going to test the AI Agent functionality, which is available in the MLJAR Studio sidebar (bottom left selection).
In the screenshot above, we see an example of the AI Agent being used. First of all, we entered the prompt asking to create a KDE plot for the petal_width variable, grouped by species. The AI agent returns a code cell that can be easily added to our notebook, by clicking Insert below active cell or Replace and Run. After executing the cell, the KDE plot we requested is shown. For additional information about the Python code cell, we can expand the AI agent thinking section, but this is optional.
AutoLab Experiments
AutoLab is a powerful feature helping us automate machine learning experiments, by using an AI agent. This lets you train and test multiple models on your chosen dataset, with minimal configuration. First of all, you need to click on AutoLab Experiments in the MLJAR Studio main menu. You are then presented with a configuration screen, where you need to specify the experiment settings, including dataset, scoring objective and various other details. Afterwards, you can automatically generate the AGENTS.md file, including all instructions and settings for the AI Agent, and finally run the experiment!
When the experiment is completed, you can see all the results, including best run trial and associated metrics. Every trial generates a separate Jupyter notebook, so you can review them to understand the experiment in detail. This helps you easily identify the optimal machine learning model for a specific objective. You can obviously save and execute the notebook of your preference, thus having reproducible results and full control!
Configuring the AI Provider
As mentioned previously, MLJAR Studio supports various types of AI models and providers. The AI provider can be configured on the AI Agent settings. The available options are listed below:
- MLJAR: This is the customized AI model provided by MLJAR. The free plan offers 50 prompts per month, but you can upgrade to pro / business and get significantly more. Click here to get additional information and plan comparisons.
- OpenAI: You can use OpenAI models by entering your API key and selecting the AI model of your preference.
- Ollama: This option lets you utilize local AI models running on your hardware. I suggest reading my article about Ollama for an extensive tutorial about installing and configuring the software.
There is significant flexibility regarding AI models, as MLJAR provides numerous options to users. You can prioritize privacy by using Ollama, or choose an AI service provider for ease of use. This is also a preferable option, if your hardware is less powerful and doesn’t support running AI models!
Modern AI tools help data scientists save time and effort, without making any compromises regarding their work output. Furthermore, business users can also utilize AI to extract data insights and analysis, while not necessarily getting involved with Python code. MLJAR provides numerous tools offering value to diverse types of users in the data analytics space. I suggest downloading MLJAR Studio and start experimenting with the powerful features it provides. You can also check my introductory book about machine learning and time series forecasting, titled Simplifying Machine Learning with PyCaret. Finally, you can follow me on LinkedIn, where I regularly post content about data science, statistics, data visualization and more!
