This post is my submission for DEV Education Track: Build Apps with Google AI Studio.
What I Built
I set out to build a full-featured Machine Learning Web App that enables users to upload CSV data, perform data exploration, advanced preprocessing, model selection (classification or regression), hyperparameter tuning, evaluation, and auto-generate Python code for deployment also generate components covering CSV upload, target selection, data exploration, preprocessing (outlier handling, datetime extraction, text handling), model and algorithm selectors, hyperparameter tuning, train-test split options, model persistence, evaluation metrics and visualizations, and final code generation.
Demo
Screenshot 1 – CSV Upload Section
Users can upload their dataset (CSV) easily with drag-and-drop or file picker, initiating the ML workflow seamlessly.
Screenshot 2 – Select Target Variable & Column to Predict
Choose the target (dependent) variable for prediction tasks, clearly distinguishing between features and labels for supervised learning.
Screenshot 3 – Data Exploration Panel
Includes:
Dataset Preview (first few rows)
- Summary Statistics (mean, median, standard deviation, min, max)
- Missing Value Detection & Visualization
- Correlation Analysis between numeric variables
Screenshot 4 – Advanced Preprocessing Options
✔️ Outlier Detection & Handling
✔️ Column Type Overrides (force categorical/numeric)
✔️ Datetime Feature Extraction (year, month, day, hour splits)
✔️ Text Column Handling (basic NLP pre-processing options)
✔️ Target Variable Transformation (e.g. log transformation for regression targets)
Screenshot 5 – Model Type Selector
🔘 Classification
🔘 Regression
Screenshot 6 – Specific Algorithm Selector
- Logistic Regression
- Random Forest Classifier
- Support Vector Machine (SVM)
- Gradient Boosting Classifier
- K-Nearest Neighbours (KNN)
- XGBoost Classifier
Screenshot 7 – Hyperparameter Tuning Panel
Interactive inputs to grid search or Random Search hyperparameters for the chosen algorithm to improve performance.
Screenshot 8 – Train/Test Split Options
✔️ Specify Train/Test Split Ratio
✔️ Enable Stratified Split (for balanced target class distribution in classification tasks)
Screenshot 9 – Model Persistence
Option to save the trained model for later inference or deployment.
Screenshot 10 – Evaluation Options Selector
✔️ Select Classification Metrics (Accuracy, Precision, Recall, F1-Score, AUC, etc.)
✔️ Choose Classification Visualizations (Confusion Matrix, ROC Curve, Precision-Recall Curve)
Screenshot 11 – Auto-generated Python Code
Displays complete, ready-to-run Python code based on all chosen parameters and configurations, enabling:
- Reproducibility
- Deployment readiness
- Educational insight for new ML engineers
🔗 Sample Generated Code: View Full Code Snippet
🌐 Live Demo: Try the App Here
💻 GitHub Repository: csv-to-python-model-generator
⭐ Contribute & Improve: Fork the repo, open issues, or start a discussion to enhance this project together!
💡 Connect with Me
👨💻 LinkedIn: Dulaj Thiwanka Jayawardena
✉️ Email: dulthiwanka2015@gmail.com
🚀 My Experience
I developed the entire project using Google AI Studio, gaining hands-on experience in Python, machine learning workflows, and integrating multiple advanced data processing features efficiently.
Top comments (0)