MLInit is a comprehensive python-notebook based framework for data wrangling and exploratory analysis.
- It provides an easy to use ipywidget based GUI Interface to input the data, label and for customizations
- Automatically Identifies the type of problem (regression/classification), type of features(numerical/categorical), clean data via regex and transforms data.
- Visualizes data in various forms for effective EDA
- Detects skewness in data distribution, performs multicollinearity checks, dimensionality reduction test and also provides missing datapoints report.
- Performs imputation based on various methods like KNN, Bfill/Ffill, iterative, simple imputation and most frequent class
- Performs stratification based on top correlated features and reports reduction in error in comparison to random split
- Performs modelling suggestion based on regression, classification problem type and allows user to tune the model hyper parameters and train.
- Reports accuracy via R2, RMSE for regression and F1, AUC for classification