Principle:Evidentlyai Evidently Text Quality Reporting

Knowledge Sources	Evidently Text Evals Evidently
Domains	NLP, Text_Analysis, Data_Quality
Last Updated	2026-02-14 12:00 GMT

Overview

A preset-based reporting mechanism that summarizes computed text descriptor statistics across a dataset.

Description

Text Quality Reporting uses the TextEvals preset to generate summary statistics for text descriptor columns. After descriptors compute row-level features (sentiment, text length, etc.), TextEvals aggregates these into dataset-level statistics: mean, standard deviation, min, max, quantiles, and value distributions.

This bridges the gap between row-level descriptor computation and dataset-level quality assessment.

Usage

Use after computing text descriptors via Dataset.from_pandas(descriptors=[...]). Include TextEvals in a Report to get aggregated statistics.

Theoretical Basis

Text quality reporting follows the compute-then-aggregate pattern:

# Step 1: Row-level computation (descriptors)
dataset["sentiment"] = [compute_sentiment(row) for row in text_column]
# Step 2: Dataset-level aggregation (TextEvals)
stats = {
    "mean_sentiment": mean(dataset["sentiment"]),
    "std_sentiment": std(dataset["sentiment"]),
    "sentiment_distribution": histogram(dataset["sentiment"]),
}

Related Pages

Implemented By

• Implementation:Evidentlyai_Evidently_TextEvals_Preset