Convert ARFF to CSV

Teams convert Weka datasets when Excel, a database importer, or a collaborator expects a flat CSV table instead of ARFF declarations. CSV is easier to inspect and exchange, but the conversion removes schema information stored in the ARFF header.

ARFF and CSV formats

ARFF (Attribute-Relation File Format) is a plain-text dataset format used primarily by the Weka machine-learning software. An ARFF file contains a header with a relation name and @ATTRIBUTE declarations, followed by an @DATA section containing records.

Attributes can be numeric, nominal, string, date, and, in Weka, relational. A nominal declaration such as @ATTRIBUTE class {yes,no} defines both the field type and its allowed values. ARFF can also store sparse rows such as {0 1,3 blue}, where omitted values use the format's default value.

CSV stores a flat table as delimiter-separated rows, normally with column names in the first row. Spreadsheet applications, pandas, R, database import tools, and many data portals accept CSV directly. CSV has no standard type declarations, nominal-value lists, relation name, comments, or sparse-row notation.

Convert with Weka

Weka is the most reliable desktop converter because it reads ARFF headers, quoted values, missing values, nominal attributes, and sparse instances.

1. Open Weka and start Explorer.
2. On the Preprocess tab, select Open file... and choose the .arff file.
3. Check the attribute list and instance count to confirm that Weka loaded the expected dataset.
4. Select Save..., choose CSV data files (*.csv), enter an output name such as dataset.csv, and save.
5. Open the result in a text editor or the intended destination program and verify the header row, delimiter, character encoding, and missing-value representation.

For repeatable conversions, run Weka's CSV saver from a terminal. Replace /path/to/weka.jar with the installed Weka JAR path:

java -cp /path/to/weka.jar weka.core.converters.CSVSaver -i input.arff -o output.csv

On Windows, quote paths containing spaces, for example "C:\Program Files\Weka-3-8\weka.jar". Use the Weka installation's supplied launcher or classpath configuration if its version requires additional libraries.

Convert with Python or an online service

pandas does not natively read ARFF files; it needs an ARFF parser. For ordinary dense ARFF files, install liac-arff and pandas:

pip install liac-arff pandas

import arff
import pandas as pd

with open("input.arff", encoding="utf-8") as source:
    dataset = arff.load(source)

columns = [name for name, _ in dataset["attributes"]]
pd.DataFrame(dataset["data"], columns=columns).to_csv("output.csv", index=False)

Test sparse ARFF files before using a Python parser in production; Weka is the safer choice when sparse instances, unusual quoting, or Weka-specific attribute definitions are involved.

For a small non-confidential file, browser services such as Zamzar and Convertio may offer ARFF-to-CSV conversion. Confirm current ARFF support and retention terms before uploading. Do not upload customer, medical, financial, research, or proprietary datasets to an external conversion service.

Compatibility checks

Preserve the original ARFF file. CSV retains values but loses declared types, relation names, attribute comments, nominal-value restrictions, and other header metadata. Keep the ARFF header or document the schema separately when recipients must distinguish numbers, dates, categories, and free text.

ARFF missing values are commonly written as ?. Depending on the converter, the CSV output may contain an empty field, ?, or another marker. Check this before importing into software that treats blanks, NA, and question marks differently.

CSV fields containing commas, quotation marks, or line breaks must be quoted according to CSV rules. Do not convert ARFF by replacing commas or splitting lines with basic text commands; those methods corrupt quoted values and cannot expand sparse rows.

Some spreadsheet installations expect a semicolon rather than a comma because of regional settings. If columns appear in one cell after opening the file, use the spreadsheet's text-import dialog and explicitly select the delimiter rather than editing the CSV manually.