heyy guysss whassupp!! so akot has happened and finally my cleaning of dataset is now very close to be perfect .. rn i just did corrected the cleaning part … just check this out and i need suggestions of how can i make it perfect
Hey guys whatsupp… hope yall are fine… so as my intelliml is coming to end …there were many errors i faced after the completion of backend during frontend dev and api integartion …. ther eare still some fixes that needs to be done but for now focusing more on better quality of eda and automl charts and reports :)
hey guyss…. so my IntelliML project is coming to its end there is still some integration left but this is what ive done so far…. the backend was already completed but frontend as well as integration of apis is on its way… for now ive integrated eda and the cleaning of dataset… check this out
hiiii…..hope yall are doin well…. so for my project…
The IntelliML backend is now functionally complete. I implemented dataset upload and versioning, profiling and quality analysis, EDA visualizations, and a full AutoML pipeline with multiple models, cross validation, hyperparameter tuning, and detailed evaluation metrics…ive also added model saving and prediction, experiment tracking and comparison, automated PDF report generation, and APIs for downloading predictions, models, and reports. I also added a dashboard summary API to track datasets, models, experiments, reports, and the best-performing model. I’m now working on improving the PDF reports with richer metadata, visualizations, and a more polished presentation.
Enhanced the AutoML Pipeline
heyyy guyss…hope yall are doin great…so
ive spent this week improving IntelliML’s AutoML pipeline by making model evaluation more detailed and reliable. Added support for precision, recall, and F1-score alongside accuracy, along with confusion matrices and classification reports for better performance analysis. Implemented 5 fold cross validation to provide more robust evaluation and added feature importance for tree-based models to improve model interpretability. I also integrated GridSearchCV for automatic hyperparameter tuning of Random Forest, measured model training time, and expanded the results to include detailed performance metrics for every trained algorithm. To make the pipeline more robust, I added exception handling so that if one model fails, the remaining models continue training without interrupting the workflow. With these improvements, IntelliML now provides a much more complete and reliable AutoML experience, and I’m now moving on to the visualization phase, started with correlation heatmaps.
##Built the AutoML Training Pipeline hii guysss………hope yall are doin well.
ive implemented the first version of IntelliML’s AutoML training pipeline, which automates the end-to-end model training process. The pipeline loads the selected dataset, handles missing values, validates the target column, performs train test splitting and one hot encoding, then trains and evaluates multiple machine learning models, including Decision Tree, Random Forest, Logistic Regression, KNN, and Gaussian Naive Bayes. It compares their performance using accuracy and automatically returns the best performing model, making model selection fast and effortless.
##Data Preprocessing Pipeline hey guyss!! hope ya’ll are doin well, so today ive build the data preprocessing pipeline for IntelliML. It can now automatically clean datasets by handling missing values, removing duplicates, encoding categorical features (Label Encoding and One Hot Encoding), scaling numerical data, and detecting outliers using the IQR method. The processed dataset is generated automatically, so it’s ready for model training without any extra manual work.
heyy guysss.. hope yall are doin well so today I focused on dataset versioning and version comparison in IntelliML. I implemented automatic version creation so every new upload is saved as a separate dataset version, along with APIs to view version history and manage stored versions. I also built the first version of the comparison module, which can compare different dataset versions by highlighting changes in rows, columns, schema, data types, missing values, and duplicate records. These features make it much easier to track how datasets evolve over time and provide a solid foundation for future experiment tracking and MLOps workflows.
I’ve implemented dataset profiling and data quality modules for IntelliMl. I’ve added APIs to analyse uploaded datasets by generating metadata(rows, cols etc).With this I’ve implemented data quality checks for missing values, duplicate rows and class distribution analysis
I’ve build a backend foundation for IntelliML in which I’ve used fastapi with a clean modular architecture. I’ve implemented dataset uploading functionality with automatic metadata storage, SQLITE, SQLAlchemy. The backend is configured with swagger api docs and I’ve started dataset profiling module to generate statistics
