Auto-Insights

• 

  Classification Performance Metric Plot (Classification Report)

• 

  Classification Dataset Upload

• 

  Regression Dataset Upload

• 

  Regression Performance Metric Plot (Residual Plot)

• 

  Pandas Profiling Report

Inspiration

The inspiration behind AutoInsights stemmed from the need to streamline the process of machine learning model development. Many data scientists and analysts face challenges in managing various tasks involved in creating, training, and evaluating models, particularly in regression and classification. AutoInsights aims to simplify and automate these processes, enabling users to focus on insights rather than technical complexities.

What it does

AutoInsights is an AutoML platform designed to facilitate classification and regression tasks. It allows users to upload their datasets, view detailed data profiles, and train and evaluate machine learning models with ease. The platform offers a user-friendly interface for selecting evaluation metrics, encoding target variables, and downloading trained models, making it accessible to both novice and experienced data practitioners.

How I built it

AutoInsights is built using Streamlit for the front-end interface, enabling an interactive and responsive user experience. The platform leverages the ydata-profiling library for generating comprehensive data profiles and provides seamless integration with custom helper functions using Pycaret libraries for regression and classification model training. The app is designed to handle various tasks through a clear and intuitive workflow, from data upload to model deployment.

Challenges I ran into

One of the main challenges I faced was ensuring the platform's flexibility to handle diverse datasets and tasks without overwhelming the user. Additionally, managing the encoding of categorical target variables in classification tasks posed a challenge, which i addressed by providing dynamic encoding options.

Accomplishments that I am proud of

I am proud of creating a comprehensive yet user-friendly platform that automates many of the tedious tasks involved in machine learning model development. The successful integration of data profiling, model training, and evaluation within a single application is a significant achievement. The ability to handle both regression and classification tasks with ease and provide downloadable trained models adds immense value for users.

What I learned

Throughout the development of AutoInsights, I learned the importance of user-centric design and the need for flexibility in handling various types of data. I also gained insights into the challenges users face in the machine learning workflow and how automation can significantly enhance productivity.

What's next for Auto-Insights

Looking ahead, I plan to expand the capabilities of AutoInsights by incorporating additional machine learning tasks, such as clustering and anomaly detection. Integrating more advanced hyperparameter tuning options and supporting more file types for dataset uploads are also on the way. Ultimately, I strive to make AutoInsights a comprehensive tool for all machine learning needs.

Built With

• pycaret
• streamlit