BDB 9.0 - Data Preparation Features & Enhancements

Welcome to the BDB Platform 9.0 Release!
In this video, we'll take you through the exciting new features and enhancements in the BDB Preparation which is a part of Data Center module.
Key updates include:
1. Auto-Generated Preparation Names: Seamlessly start with auto-generated names like "Prep 1", "Prep 2", etc.
2. Integrated Profiling Chart: Access column profiling data directly from the column menu.
3. Dataset-Level Suggestions: Get immediate insights to streamline your workflow.
4. Source and Sample Size Integration: View comprehensive dataset details in one place.
5. Drawer Interface for Transforms: Simplify transform operations with a new user-friendly drawer interface.
6. User Feedback Snack Bars: Quickly identify errors and successes with dismissible snack bars.
7. Consolidated Data Profile Tab: Analyze data comprehensively with integrated Profile, Chart, Suggestion, and Pattern info.
8. Automatic Deactivation of Auto Prep: Ensure single-use functionality for the Auto Prep feature.
9. New Save and Close Button: Enjoy easy navigation and automatic saving of steps.
Additionally, we dive into various new transformations to improve data analysis and machine learning tasks:
1. Binning/Discretize Values: Convert continuous data into distinct categories.
2. Expanding Window Transform: Generate new features by computing rolling statistics over expanding windows.
3. Feature Agglomeration: Reduce redundancy in datasets with numerous features.
4. Label Encoding: Convert categorical columns into numerical ones for ML compatibility.
5. Lag Transform: Shift or delay a time series by a certain number of time units.
6. Leave One Out Encoding: Encode categorical variables based on the target variable without causing data leakage.
7. One Hot Encoding: Create new columns for each unique value in a column.
8. Principal Component Analysis (PCA): Reduce dimensionality and identify patterns in numerical columns.
9. Rolling Data: Create new features via computations within sliding windows for time series analysis.
10. Singular Value Decomposition (SVD): Decompose a matrix for data compression, noise reduction, and feature extraction.
11. Target Based Quantile Encoding: Encode categorical variables considering the target variable distribution for regression tasks.
12. Target Encoding: Replace categories with target variable means to enhance ML model training.
13. Weight of Evidence Encoding: Encode categorical variables based on their predictive power for binary classification.
14. Formula-Based Transforms: Apply specific calculations or express relationships between variables for preprocessing and feature engineering.
Thank you for watching the video.
We hope it brings clarity to your data preparation process!
For more details please visit our website www.bdb.ai and refer to our documentation section.