javascriptdata/danfojs
Danfojs
Danfo.js is a data analysis and preprocessing library for JavaScript that provides high-performance labeled data structures. It implements data frames and series to enable complex data analysis, statistical computing, and the manipulation of structured tabular data.
The project serves as a machine learning preprocessing library, offering utilities for categorical label encoding, one-hot encoding, and numeric feature scaling and standardization. It specifically facilitates the conversion of labeled data structures into tensors for model training and evaluation.
The library covers a broad set of capabilities including descriptive statistics, relational operations like merging and joining, and time-series processing. It includes tools for data cleaning, filtering, and grouping, as well as a visualization interface for generating interactive charts and plots directly from data frames.
The system supports importing and exporting data via CSV, JSON, and Excel formats.
Features
- JavaScript Data Transformations - Implements a comprehensive framework for processing and transforming structured tabular data using dataframes and series in JavaScript.
- Tabular Data Structures - Implements high-performance labeled data structures for managing relational data with column mutation and label alignment.
- Tensor Conversion Utilities - Transforms structured labeled data into tensor formats for use in machine learning workflows.
- Data Preprocessing for Modeling - Offers specialized utilities for cleaning, scaling, and encoding datasets to prepare them for machine learning model training.
- Data Analysis Libraries - Serves as a high-performance library for cleaning and transforming structured datasets within JavaScript environments.
- Data Cleaning Procedures - Provides procedures for cleaning datasets, including the removal of duplicate values and noise filtering.
- Storage - Uses TensorFlow.js tensors as the underlying storage mechanism to enable high-performance vectorized operations.
- Data Concatenations - Combines multiple data frames or series into a single object along a specified axis.
- Data Grouping Utilities - Provides utilities for organizing data into groups based on keys or labels for aggregated calculations.
- Data Joins - Merges and joins multiple data structures together based on shared keys or indices.
- Tensor Transformations - Provides utilities to transform structured data frames into tensors for compatibility with machine learning frameworks.
- Labeled Data Structures - Implements high-performance labeled data structures for complex statistical computing and data analysis.
- Data Subsetting and Extraction - Retrieves specific columns, rows, or cross-sections using labels, integer positions, or slice notation.
- Exploratory Data Analysis - Provides tools for calculating descriptive statistics and generating charts to discover patterns in datasets.
- Group-By Aggregations - Provides core group-by operations to perform summary calculations and extract insights from relational datasets.
- Label-Based Data Selection - Implements label-based indexing to allow data retrieval and alignment using descriptive row and column identifiers.
- Labeled Series Data Structures - Stores labeled data in a series format with support for custom indices and data types.
- Missing Data Removal - Identifies and handles null, undefined, or NaN values to ensure data quality.
- Position-Based Data Selection - Retrieves specific data subsets using integer indices, arrays of positions, or slice notation.
- Group Extractions - Implements the ability to retrieve a standalone data structure containing only rows associated with a specific group name.
- Tabular Data Managers - Supports importing, merging, and joining datasets from CSV, JSON, and Excel into unified relational structures.
- Tabular Data Organization - Enables cleaning, filtering, and aggregating of relational datasets using column-based operations.
- Tabular Structure Creation - Builds dataframes and series from JSON objects, arrays, tensors, and dictionaries.
- Inter-Format Conversions - Converts between arrays, JSON, lists, objects, and tensors to enable interoperability between formats.
- Element-wise Comparisons - Evaluates logical relationships between data series using element-wise comparisons and broadcasting.
- Element-wise Array Operations - Executes high-performance mathematical functions across entire data series simultaneously using tensor acceleration.
- Statistical Analysis Libraries - Calculates essential descriptive statistics including mean, median, variance, and standard deviation.
- Feature Scale Normalization - Transforms numeric features to fit within specified ranges to ensure stable model convergence.
- One-Hot - Converts categorical labels into one-hot numeric arrays for machine learning tasks.
- Cumulative Aggregate Calculations - Computes running totals, products, and extrema along specific data axes.
- Categorical Encodings - Converts categorical text labels into unique integers for use in numerical machine learning models.
- Data Alignments - Synchronizes datasets using automatic label-based alignment during mathematical computations.
- Statistical Standardization - Rescales numeric data to a mean of zero and a standard deviation of one.
- Interactive Data Visualizations - Implements interactive line and box plots allowing users to visually explore data frames and series.
- Data Type Casting - Converts column data types to ensure consistency for processing and analysis.
- Data Export - Writes structured data frames or series into CSV, Excel, or JSON files for external storage.
- Data Visualization Interfaces - Provides a customizable interface for rendering and analyzing structured data via interactive plots.
- Dataset Label Management - Provides tools to modify axis labels, reset indices, and remove specified rows or columns from datasets.
- Expression-Based Data Querying - Extracts subsets of data by evaluating boolean expressions against the columns of a dataset.
- Series Value Mappings - Applies custom functions or mapping correspondences to transform every value within a data series.
- Tensor Conversions - Transforms labeled data structures into tensors required for training and evaluating machine learning models.
- Grouped Cumulative Metrics - Calculates running totals, products, or extrema for each group to track progress.
- Split-Apply-Combine Patterns - Implements the split-apply-combine pattern to segment datasets, apply functions, and reassemble results.
- Tabular Data Import - Reads relational data from CSV, Excel, and JSON files into structured formats for analysis.
- Tabular Data Sorting - Rearranges data frame rows based on the values of specific columns in ascending or descending order.
- Format-Agnostic Converters - Converts diverse input formats such as JSON and CSV into internal typed tabular structures.
- Time Series Analysis - Includes analytical methods for processing date-based indexing and managing sequential temporal data.
- Unique Value Counting - Identifies unique elements and counts their occurrences to determine the mode of the dataset.
- User-Defined Functions - Allows the application of user-defined functions across data axes to transform or analyze values.
- Boolean Masking - Creates boolean masks via logical expressions to filter and index specific data subsets.
- Chart Generators - Provides utilities to render various graphical representations like line, bar, and scatter plots from structured data.
- Descriptive Statistics Summaries - Generates a bundled statistical summary to provide an overview of the dataset's distribution.
- Data Trend Visualizations - Generates a variety of charts to identify patterns and trends within tabular datasets.
