Analyze CSV Data

Who This Page Is For

• Users who want quick CSV statistics from the command line
• Analysts preparing CSV data for charts, reports, observers, or inference
• Developers checking the current analyze-csv parser and pipeline behavior

Quick Start

indexly analyze-csv sales.csv --show-summary

Add cleaning and a terminal histogram:

indexly analyze-csv sales.csv \
  --auto-clean \
  --show-summary \
  --show-chart ascii \
  --chart-type hist \
  --transform auto

For mixed file folders, use the universal dispatcher:

indexly analyze-file sales.csv --auto-clean --show-summary

What analyze-csv Does

The CSV pipeline runs in this order:

1. Detect the delimiter from common delimiters such as comma, semicolon, tab, pipe, colon, and tilde.
2. Load the CSV with UTF-8 handling.
3. Optionally run the cleaning pipeline when --auto-clean is set.
4. Infer numeric columns when most values in a text column can be converted.
5. Compute statistics for numeric columns or derived timestamp columns.
6. Optionally render charts or export analysis output.
7. Persist analysis results unless --no-persist is set.

Statistics Produced

Indexly computes these statistics for each numeric column:

If a cleaned CSV only contains datetime values, derived _timestamp fields can provide numeric columns for analysis.

Visualization Options

Use --show-chart to choose where the chart renders:

Supported chart types:

--chart-type bar
--chart-type line
--chart-type box
--chart-type hist
--chart-type scatter
--chart-type pie

Common chart controls:

--x-col date
--y-col revenue profit
--export-plot chart.html
--agg sum

For histogram and boxplot distribution work, transformation can improve readability:

--transform none
--transform log
--transform sqrt
--transform softplus
--transform exp-log
--transform auto

--transform auto chooses a transformation from column skew. ASCII histogram bars use --bar-scale sqrt by default and also accept --bar-scale log.

Time-Series Analysis

For date-indexed CSVs:

indexly analyze-csv sales.csv \
  --auto-clean \
  --timeseries \
  --x order_date \
  --y revenue,profit \
  --freq M \
  --agg sum \
  --rolling 3 \
  --mode interactive \
  --output sales-trend.html

See Time-Series Visualization for dedicated examples.

Boxplot Engine

For focused boxplot work, use the isolated boxplot engine:

indexly analyze-csv sales.csv \
  --boxplot \
  --group-by region \
  --y-col revenue profit \
  --show-mean

Useful boxplot options include:

For analytical artifact lifecycle:

• Default re-analysis behavior prunes superseded hash-versioned Parquet artifacts.
• --keep-artifact-history preserves old artifacts.
• Use indexly clear-data --all --prune-artifacts to remove unreferenced artifact history later.

Export Results

Export the analysis table:

indexly analyze-csv sales.csv --export-path reports/sales.md --format md

Supported analyze-csv formats are:

• txt
• md
• json

Use --compress-export with JSON output when you want .json.gz:

indexly analyze-csv sales.csv --export-path reports/sales.json --format json --compress-export

--export-format is accepted as an alias for CSV analysis output format. For broader tabular export formats such as CSV, Parquet, Excel, or SQLite, use indexly analyze-file with --format.

Practical Workflow

indexly rename-file ./exports --pattern "{date}-{title}" --recursive --dry-run
indexly analyze-csv ./exports/sales.csv --auto-clean --show-summary
indexly analyze-csv ./exports/sales.csv --show-chart ascii --chart-type hist --transform auto
indexly observe audit

This keeps filenames stable, cleans and analyzes the CSV, then lets observers compare persisted snapshots when observer data is available.

Related Pages

• Data Analysis Overview
• Clean CSV Data
• Time-Series Visualization
• CSV Inference
• Rename File