Compute Correlation Matrix for Parquet Files Online
Compute a Pearson correlation matrix for numeric columns in Parquet files directly in your browser. Instantly spot which variables move together — colour-coded heatmap, no upload required.
How to compute Correlation Matrix for Parquet files
- Drop your file onto the upload area. Every numeric column is found and selected.
- Click column names to leave out any you do not want, such as ID or ZIP code columns. At least two must stay selected.
- Click Compute Correlations. A colour-coded matrix appears, blue for positive and red for negative, with values to 3 decimals. Hover a cell for 6 decimals.
- Click Export Matrix CSV to download the matrix with values to 4 decimals.
Your file is processed locally in your browser and is never uploaded. The free limit is 50 MB per file; larger files work if your device has the memory for them.
Worked example
An estate agent has recent house sales and wants to see which property features move with the sale price before building a pricing model.
Input (Parquet)
Parquet file (binary, columnar) — shown as a table with its schema
| floor_area_sqft | bedrooms | age_years | sale_price |
|---|---|---|---|
| 1450 | 3 | 32 | 612000 |
| 2100 | 4 | 8 | 845000 |
| 980 | 2 | 55 | 455000 |
| 1720 | 3 | 20 | 701000 |
| 2600 | 5 | 3 | 990000 |
| 1200 | 2 | 41 | 540000 |
Schema: floor_area_sqft BIGINT, bedrooms BIGINT, age_years BIGINT, sale_price BIGINT
Settings
- Numeric columns: floor_area_sqft, bedrooms, age_years, sale_price (all selected)
- Export: Export Matrix CSV
Result
| floor_area_sqft | bedrooms | age_years | sale_price | |
|---|---|---|---|---|
| floor_area_sqft | 1.0000 | 0.9830 | -0.9669 | 0.9991 |
| bedrooms | 0.9830 | 1.0000 | -0.9306 | 0.9798 |
| age_years | -0.9669 | -0.9306 | 1.0000 | -0.9721 |
| sale_price | 0.9991 | 0.9798 | -0.9721 | 1.0000 |
The output is always a CSV, whatever the input format, with an empty top-left header cell. Floor area tracks price almost perfectly at 0.9991. Age is strongly negative at -0.9721: older homes sold for less. The diagonal is 1 because each column matches itself, and the matrix is symmetric. With only six rows these values are fragile, and one unusual sale could move them a lot.
Working with Parquet files
Integer, float, double and DECIMAL columns are all picked up from the Parquet schema, so there is no inference to go wrong. That also means numeric columns you would not want are selected too, such as surrogate keys, year numbers or store IDs. Deselect them before computing. They add rows and columns to the matrix that look precise and mean nothing. An auto-increment ID, for example, often shows a strong correlation with any value that grew over time.
The engine computes every pair in one scan of the table, which keeps wide files manageable. Twenty columns means 400 correlations from a single pass. Boolean columns and timestamps are not offered. If you want the effect of a flag, store it as 0 and 1 in an integer column. The result is not written back to Parquet. You get the on-screen matrix and a CSV download of the coefficients.
Frequently Asked Questions
Can I download the correlation matrix as Parquet?
No. The matrix is exported as CSV only. Convert that CSV to Parquet with the converter if you need it.
Are DECIMAL columns in my Parquet file included?
Yes. DECIMAL, integer, float and double columns are all offered.
Which correlation method is used?
Pearson correlation, which measures straight-line relationships. A strong curved relationship can still show a value near 0. Spearman and Kendall are not available.
Why does a cell show NaN or a dash?
NaN means one of the two columns has the same value in every row it shares with the other, or only one shared row exists, so there is no variation to correlate. A dash means there were no rows with values in both columns. In the CSV export a dash becomes an empty cell.
Is there a limit on the number of columns?
There is no fixed limit, but the matrix grows with the square of the column count. Deselect IDs and other columns that are not real measures to keep it readable.
Related Tools
Detect Outliers in Parquet Files Online
Detect statistical outliers in Parquet files directly in your browser. Flag or remove rows where numeric values exceed a chosen number of standard deviations from the mean — no upload required.
Normalize Columns in Parquet Files Online
Normalize numeric columns in Parquet files using min-max scaling (0–1) or z-score standardisation (mean=0, std=1). Adds new columns alongside the originals — no upload required.
Manage Columns in Parquet Files Online
Drop or select specific columns from Parquet files directly in your browser. No upload required.
Compute Correlation Matrix for CSV Files Online
Compute a Pearson correlation matrix for numeric columns in CSV files directly in your browser. Instantly spot which variables move together — colour-coded heatmap, no upload required.
Compute Correlation Matrix for Excel Files Online
Compute a Pearson correlation matrix for numeric columns in Excel files directly in your browser. Instantly spot which variables move together — colour-coded heatmap, no upload required.
Compute Correlation Matrix for JSON Files Online
Compute a Pearson correlation matrix for numeric columns in JSON files directly in your browser. Instantly spot which variables move together — colour-coded heatmap, no upload required.
Parquet Viewer Online
View and inspect Parquet files directly in your browser. Browse rows, check column names and data types — no upload required, your data stays on your device.
Convert Parquet to CSV Online
Convert Parquet files to CSV format directly in your browser. No upload required — your data never leaves your device.