Several commnads are here to customize the parsing of the CSV file. Each command takes one string parameter.

• \delimiter indicates the delimiter that separates columns.
• \separator indicates the decimal separator
• \thousands indicates the thousands separator
• \quotes indicates the quotes separator

Every chart and every calculation uses the columns that are currently shown, in the
order they are shown, and only the rows that currently match your search.

So, before drawing a chart or asking for statistics:

1. Hide the columns you do not need.
2. Drag the column headings to put them in the order you want.
3. Filter the rows if you only care about some of them.

This is how you tell a chart which column goes on which axis. There is no separate
setup screen: the table itself is the setup.

• Search: type in the search box. Only rows containing what you typed remain.
• Sort: click a column heading to sort by it. Click again to reverse the order.
• Pages: Show … entries chooses how many rows per page. Use the page numbers at the bottom right to move around.
• Wide tables: scroll sideways.
• Your layout is remembered: the column order, the hidden columns, the sort and the search are kept in this browser, so you find the table as you left it.
  Another browser or another computer starts from the original layout.
• Click on a cell, and something useful happens depending on what it contains:

The search bar displays the rows for which there is a match.

• Column visibility lists all columns with a tick for each one that is shown.Untick the ones you do not need.
• Reorder columns by dragging a column heading left or right.

• Search Panes shows, for some columns, the list of their values with counts.

  Click a value (for example Paris) to keep only those rows. Click several to keep several.
• Search Builder lets you combine conditions, for example *Temperature greater than 30 and City equals Lyon*.
• The plain Search box above the table filters on any column at once.

Rows that you filter out are not lost. On the charts below they appear in grey, so you can see them in context.

Copy, CSV, Excel and PDF follow the standard behaviour of these buttons. The two "visible" buttons follow the golden rule above, so they are the way to keep a filtered, trimmed-down version of a table.

Sometimes the number you need is not in the table, but can be calculated from it: a difference, a ratio, a unit conversion. Derived column creates a new table with an extra column computed for every row.

1. Show the columns you want to keep in the new table, and filter the rows if needed.
2. Click Derived column. A window opens.
3. Write the expression, or click the column names to insert them. Column names go in square brackets: [Temperature].
4. As you type, the window tells you whether the formula works, how many rows get a value, and shows the first few results.
5. Give the new column a name and the new CSV a name, then click Submit.

The new CSV contains your shown columns plus the new one. As with New CSV from visible, its file name is copied to your clipboard.

What you can write

Examples on the Daily weather table

For Gap, the window previews the first values: 0.6, 1.4, 1.8, −1.6, 1.4 …

Good to know

• Only numeric columns can be used in a formula. A column of words or dates cannot.
• A row gets a blank value when one of the numbers it needs is missing, or when the result is impossible (dividing by zero, square root of a negative number). The window tells you how many rows are blank. If no row gets a value, it tells you so and does not let you submit.
• The new column name must be different from the existing column names.
• Calculations follow the usual order: powers first, then × and ÷, then + and −. Use parentheses when in doubt.

Click a chart button below the table. The chart appears under the table, built from the shown columns, in their shown order.

If the columns do not fit (for example a column of words where numbers are needed, or too few columns), the chart is simply not drawn. Check the shown columns and their order.

The fitted line on the scatter chart. Besides the dots, a dashed line shows the straight line that best follows them. Its label gives the equation and an R² value, for example:

Linear fit: y = 1.313x − 7.594, R² = 0.993

Read it like this: for each extra degree on the horizontal axis, the vertical value goes up by about 1.31. R² tells how well the line fits, between 0 and 1: close to 1 means the dots lie almost on the line; close to 0 means the line explains little. Click the line’s name in the legend to hide it. When the horizontal axis is made of dates, the label shows only R².

Below the chart you will find:

• Hide Chart: removes the chart.
• The download button: saves the chart as an image.
• Gaussian σ: a slider that smooths line charts and 3D surfaces. At 0 you see the raw data; moving it right averages neighbouring points to reveal the general shape. The max box sets the end of the slider.

2.5.1 Surface visualization

2.5.2 Plot smoothing

2.5.3 Statistics and analysis

2.5.4 The math behind statistics

Here is a sample weather data table to work with.

Question: on the Daily weather table, how well does the thermometer reading predict how hot it feels, and is humidity involved?

1. Choose the columns. Open Column visibility and hide Date and City. Drag Apparent temperature right after Temperature, so the order is Temperature, Apparent temperature, Humidity, Rain.
2. Draw a scatter chart. Click Scatter Chart. Temperature is on the horizontal axis, Apparent temperature on the vertical axis. The dots lie almost on a line, and its label reads y = 1.313x − 7.594, R² = 0.993: the feeling follows the thermometer very closely, and rises faster than it.
3. Look at all the relations at once. Click Correlation Matrix. The Temperature–Humidity cell is red (about −0.81): humid days are cooler.
4. Zoom on one city. Show the City column again, type Paris in the search box, and click Statistics: the figures now describe Paris only.
5. Measure the gap. Clear the search box so that all cities are back, then click Derived column, write [Apparent temperature] - [Temperature], name it Gap, name the CSV weather_gap, and submit. You now have a new table where you can sort by Gap to find the days when the feeling differed most from the thermometer.

What follows is an excerpt from the dataset Complete Historical Cryptocurrency Financial Data .