DataPrism is a local-first Exploratory Data Analysis (EDA) tool. Drop in any CSV or JSON file and immediately audit data quality, outliers, missing fields, and correlation matrices right in your browser or inside VS Code.
Drop a CSV or JSON file in the mockup below to run our full analysis engine entirely client-side. No data is ever uploaded to any server.
DataPrism incorporates a high-fidelity diagnostic engine running fully client-side to summarize, clean, and enrich datasets.
Computes Fisher-Pearson skewness, standard deviation, and quartile interpolations client-side with zero latency.
Applies Tukey's fences (IQR bounds) to identify anomalies, list indexes, and score overall dataset cleanliness.
Flags target leakage, recommends scaling/log transforms, and writes copyable Python pandas recipes for data preparation.
Builds full Spearman and Pearson correlation grids and renders interactive heatmap visualizations natively inside your IDE.
DataPrism conducts all numerical computations locally. Key algorithms include:
$$\sigma = \sqrt{\frac{1}{N} \sum_{i=1}^{N} (x_i - \mu)^2}$$
Quantifies the variance of distribution values around the arithmetic mean $\mu$.
$$g_1 = \frac{\frac{1}{N} \sum_{i=1}^{N} (x_i - \mu)^3}{\sigma^3}$$
Determines asymmetry. If skewness is high, DataPrism recommends log transforms.
$$[Q_{25} - 1.5 \times \text{IQR},\, Q_{75} + 1.5 \times \text{IQR}]$$
Values outside these boundaries are identified as anomalies and recorded in the audit logs.
DataPrism runs completely offline as a standard extension. Add it to your editor workspace now.
ext install CODExGAMERZ.dataprism