Data

The profile
The Data table
Argument axes and units
Value axes
The logarithmic axis
A custom argument axis
Working on part of the profile
Smoothing the profile
Data sources
Sample data
Web address
Published data (DOI)

The profile

This software's choice — A decision made by this software rather than by the field's sources; the explanation says why it was made.

The profile is the measured data a model is fitted to: a list of points, each an argument and the amplitude measured at it.

When you import a data file (File > Import Profile) or open a project, its points become the profile. Each point has two numbers: the argument, which runs along the horizontal axis - an angle, an energy, a time or any other position - and the amplitude, the value measured there.

The profile is drawn on the chart and listed in the Data tab on the left, beside the Tools tab. Everything the program proposes or fits is computed from it: the curve positions it finds, the fit intervals, and the difference between the model and the data.

The argument is stored exactly as it was read. What the horizontal axis shows - its name, its unit and its numbers - is chosen separately under Data > Argument Transformation, and changing it never changes the stored values or the fit.

The Data menu works on the profile itself: it chooses how the argument is shown, narrows the work to a stretch of the profile (Data > Range) and smooths the amplitudes (Data > Smooth Profile). The tables at the bottom of the window - Fit Intervals, Curve Positions and Curve Attributes among them - describe the model rather than the data.

Almost every command of the Data and Model menus is greyed until a profile is loaded, and while a fit or another calculation is running.

Limitations

Smoothing, editing a point and anything a module subtracts from the profile all change the profile the program holds, not the data file. File > Reload Profile reads the file again, which is the way back to the data as measured.

The Data table

This software's choice — A decision made by this software rather than by the field's sources; the explanation says why it was made.

The Data tab on the left lists every point of the profile, or of the data interval in force, and lets you correct a value by typing over it.

The table has two columns: Position, the argument, and Amplitude, the value measured there. While a data interval is selected (Data > Range) it lists only the points of that interval.

To correct a point, click its cell, type the new number and leave the cell. The point is replaced in the profile the engine fits, and the chart is redrawn from it.

The tab shares the left panel with the Tools tab; click its tab to bring it forward.

A cell holding anything that is not a number - a typo, a stray letter, a unit typed after the value - is refused: a message beside the table names what was typed, and the point stays where it was.

Limitations

An edit changes the profile the program holds, not the data file. Save the project to keep it; File > Reload Profile discards it.

Argument axes and units

This software's choice — A decision made by this software rather than by the field's sources; the explanation says why it was made.

The horizontal axis can show the stored argument in several ways - as the model and the data say it should, as a plain position, on a logarithmic scale, through formulas of your own or as a module included in the build reads it - without changing the data or the fit.

The program stores the argument of every point exactly as it was loaded. An axis rule is a display transform: it decides the name and unit written under the chart, the numbers along the axis, the caption and reading of the pointer's position in the panel on the right, and how positions read in the Curve Attributes table. The stored data and the fit are the same whichever rule is in force.

Choose a rule under Data > Argument Transformation > Use Rule, or right-click the chart and choose it under Argument in the menu that opens there. Both offer the same entries, and the one in force is ticked in both:

Data > Argument Transformation > Use Rule > Automatic - what the model and the data say the argument is. The program asks, in order: the curves the model holds, when they agree; then the data, whose reader says what its argument is; then what the model's curves assume when the data says nothing; then, only while the model is empty, the curve type selected on the Tools tab. When none of them says, the argument is shown as Position.
Data > Argument Transformation > Use Rule > General Position - the argument as loaded, labelled Position, with no unit.
Data > Argument Transformation > Use Rule > Logarithmic - the argument on a logarithmic scale, marked and read in the argument itself. See the logarithmic axis.
Data > Argument Transformation > Use Rule > Custom Position - an axis defined by a formula of your own and its inverse.

A module can add entries of its own to this menu; its own chapter of the guide says what they are.

Until you pick a rule yourself the program uses Automatic, so the axis follows the model and the data. Once you pick any rule, the choice is yours, and neither the model nor the data changes it. Choose Automatic to hand the axis back.

The choice belongs to the project, not to the program: it is saved with the project and comes back when the project is reopened, and File > Save Project counts it as a change to save. A new project, and a different data file imported into the one open, start on Automatic again, so one project's axis never follows you into the next.

Use Rule is greyed while nothing is drawn on the chart. The axis label always reads as the name followed by the unit in square brackets, or the name alone when the rule has no unit; the caption beside the pointer's position on the right is the same name.

Limitations

A rule saved in a project that cannot be used - one missing a number it needs, a custom axis without both of its formulas, or an entry a module added to a build that no longer contains that module - is replaced by Automatic while the project is open.
Every entry is offered whatever the data is. The program cannot tell whether your argument really is an angle or a date, and on data that is not, those entries show numbers that mean nothing.

Value axes

This software's choice — A decision made by this software rather than by the field's sources; the explanation says why it was made.

The vertical axis is chosen the same way as the horizontal one: what the model and the data say the value is, a plain value, a logarithmic scale or formulas of your own - without changing the data or the fit.

The value of every point is stored exactly as it was loaded. A value rule decides the name written beside the vertical axis, the numbers along it, where each point is drawn, and the caption and reading of the pointer's value in the panel on the right.

Choose a rule under Data > Value Transformation > Use Rule, or right-click the chart and choose it under Value:

Data > Value Transformation > Use Rule > Automatic - what the model and the data say the value is, asked in the same order as for the argument. When nothing says, the value is shown as Value.
Data > Value Transformation > Use Rule > General Value - the value as loaded, labelled Value.
Data > Value Transformation > Use Rule > Logarithmic - a logarithmic chart. See the logarithmic axis.
Data > Value Transformation > Use Rule > Custom Value - an axis defined by a formula of your own and its inverse, in the same dialog as a custom argument axis.

A module can add entries of its own to this menu too; its own chapter of the guide says what they are.

As with the argument, a rule you pick is saved with the project, and Automatic hands the choice back.

Limitations

The rule moves the drawn points and the readout, not the tables: the Data and Curve Positions tables, and a curve's Amplitude in Curve Attributes, show the stored values. An amplitude is a height above whatever lies beneath it, not a point on the vertical axis, so a logarithm of it would not be where the curve is drawn.
The fit is always made against the stored values. A logarithmic chart shows the data differently; it does not fit its logarithm.

The logarithmic axis

Convention — Common practice or a guideline in the field, not a rule its sources require.

Logarithmic draws a coordinate at the decimal logarithm of its value, so equal ratios are equal distances, and marks and reads it in the value itself.

It is offered for both axes: Data > Argument Transformation > Use Rule > Logarithmic and Data > Value Transformation > Use Rule > Logarithmic. On the value axis it is the ordinary logarithmic chart of prices, counts or concentrations, on which a steady percentage growth is a straight line; on the argument axis it is how a dose or a concentration series is usually plotted.

The axis keeps the quantity's own name and adds ", log scale" - Value, log scale. Its marks are the values themselves (100, 316.2, 1000), not their logarithms, and so is the reading of the pointer's position on the right.

Limitations

Zero and negative numbers have no logarithm. A point whose value is zero or below is left out of the chart - its line stops on either side of it and the pointer does not snap to it - while it stays in the data and in the fit. On the value axis the Difference curve, which is negative wherever the model lies above the data, is therefore drawn only where it is positive.
The spacing between marks is chosen in the logarithm, so over less than one decade the marks fall at values such as 112.2 and 125.9 rather than at round numbers.

A custom argument axis

This software's choice — A decision made by this software rather than by the field's sources; the explanation says why it was made.

Data > Argument Transformation > Use Rule > Custom Position shows the argument through a formula of your own, given together with its inverse.

The entry opens the Custom Argument Axis dialog, which has four fields:

Display name and Unit - the label under the axis, written as the name followed by the unit in square brackets. The unit may be left empty.
Displayed value f(x) - the value shown, as a formula of the stored value x.
Inverse g(x) - the way back, from a shown value to the stored one. It is needed because the chart converts in both directions: reading a position off the axis is the inverse of drawing it.

For a natural-logarithm axis, write ln(x) as f(x) and exp(x) as g(x). The formulas are written as for a user-defined curve: ln is the natural logarithm, log is base 10, and exp, sqrt, abs, sin, cos and the operators + - * / ^ can be used with brackets.

The first time, the dialog is filled in with the name Custom, no unit, and x for both formulas - the axis you already have - so pressing OK unchanged changes nothing. After that it opens on the definition you last gave, which is saved with the project.

If either formula is left empty, a message says both are required and the axis stays as it was.

The value axis has one too, at Data > Value Transformation > Use Rule > Custom Value, which opens the same dialog for the vertical axis. The two definitions are kept apart.

Limitations

The program does not check that g(x) really is the inverse of f(x). If it is not, values read back from the chart land in the wrong place, and nothing warns you.
Like every axis rule, a custom axis changes only what is shown; the fit works on the stored values.

Working on part of the profile

This software's choice — A decision made by this software rather than by the field's sources; the explanation says why it was made.

Data > Range narrows the program's work to the stretch of the profile between two points you pick, and brings the whole profile back.

Choose Data > Range > Start Manual Selection and click two points of the profile on the chart, one at each end of the stretch you want. The entry is ticked while this picking mode runs, the status line says which click comes next, and the picks are drawn as the Area Limits series. Until the second pick, clicking the first again takes it back; once both are picked, further clicks are ignored.

Then choose Data > Range > Select Data Interval. It is greyed until exactly two points are picked. The profile is cut to the points between them, both ends included; the chart and the Data table show only that stretch, and the picking mode ends.

From then on the automatic commands - proposing curve positions and fit intervals, and those a module adds - work on that interval only, and so does the fit.

Data > Range > Select Entire Profile goes back to the whole profile. It is offered only while an interval is in force and no picking mode is running.

A selected interval is saved with the project, as the numbers of its first and last points.

Selecting another interval while one is in force replaces it: the fit runs over the new one.

Smoothing the profile

This software's choice — A decision made by this software rather than by the field's sources; the explanation says why it was made.

Data > Smooth Profile replaces each amplitude by the mean of itself and its two neighbours, keeping the total of all amplitudes the same.

Each point's amplitude becomes the mean of three values: the point before it, the point itself and the point after it. At either end the missing neighbour is replaced by the end point itself. The smoothed amplitudes are then scaled so that their sum equals the sum before smoothing: the area under the profile is kept, and the highest peaks come down a little.

Use it on noisy data before the automatic commands, which look for local maxima and are easily misled by noise. It can be applied more than once; each application smooths further.

The entry is offered whenever a profile is loaded and no fit is running. The chart and the Data table show the smoothed profile at once.

Limitations

Smoothing changes the data you fit, and a fit to a smoothed profile describes the smoothed data rather than the measurement. It cannot be undone except by File > Reload Profile.
It smooths the whole profile. A data interval selected before smoothing keeps its values as they were, so smooth first and then select the interval.

Data sources

This software's choice — A decision made by this software rather than by the field's sources; the explanation says why it was made.

File > New Project from Data Source finds data in a published collection or at an address, downloads it and starts a project with it.

The window has the sources down its left-hand side, grouped by what they hold; a module adds its own, so the list depends on which modules were built in. They stay there the whole time, so another source is one click away at any point. Choosing one fills the right-hand side with what that source needs: what to look for, which file of a record to take, and a preview of the data before anything is imported.

The line of steps above that panel says which of them you are on and what is still to come. A step already answered shows the answer beside its name and can be clicked to go back to it, as can the Back button; a step not yet reached is greyed, and clicking it says what has to be finished first.

Nothing has to be looked up first. Where a source knows what it holds, it offers it by name and the identifier it uses behind the scenes is never shown; where a service knows too much to list - every substance, every published dataset - the common answers are offered and anything else can be typed. Dates are chosen from a calendar rather than typed in a format to get wrong.

A question that matters only for one answer to another is greyed until that answer is chosen, and the line under it says which - dates, say, are asked only when a range is chosen as custom dates rather than as the whole series or the last few years. What is left in a greyed box is not sent.

While a file is being fetched the window keeps working: the line at the bottom shows what has arrived and how long it has taken, and Cancel becomes Stop, which gives up on the download and leaves the wizard as it was.

The preview step says where the file was saved and lets you change it: Save in shows the folder, and Change moves what was just fetched into another one. The choice is remembered, so it is a row to read rather than a question to answer every time. By default it is a folder of downloaded data under your user account.

The project is then started from that file exactly as File > Import Profile starts one from a file you already have. It records where the data came from - the source, what was asked for, the address and when it was fetched - and stores the data itself, so reopening it never needs the network. The project itself is not saved until you save it, with File > Save Project.

The preview is read by the same reader that will import the file, so what the preview shows is what the project gets. A file whose kind this build has no reader for is shown greyed, with the reason.

Looking is free of consequences. Every file the wizard fetches is written to disk so that it can be previewed, and the ones that did not become a project are thrown away when Fit closes - so trying six series leaves one file, not six. A file kept in a folder you chose yourself is never deleted, and neither is one a project was made from: that project reads it again when you ask for File > Reload Profile.

Limitations

Fetching data needs a network connection and the curl program: every macOS and Windows 10 (version 1803) or later includes it, and the Linux packages install it. Downloads use the system's own secure connections and certificates, and a proxy set in the http_proxy or https_proxy environment variable. Fitting itself never needs the network: a project, once saved, holds its data.
Fit fetches data and nothing else. It does not search for a matching phase, identify a substance or interpret what it downloads.
Each service decides what it allows and how often it may be asked. The terms shown beside a source are what its publisher states; a download is the user's own use of that service.

Sample data

This software's choice — A decision made by this software rather than by the field's sources; the explanation says why it was made.

The Sample data source offers the example files installed with Fit, so there is something to fit before any data has been found.

It lists the files in Fit's own data directory that this build has a reader for - the examples its analysis modules installed. Nothing is downloaded and no connection is used.

They are ordinary files: the same profiles can be opened with File > Import Profile.

Limitations

An installation that did not install the sample directory has nothing to offer here, and the source says so rather than showing an empty list.

Web address

This software's choice — A decision made by this software rather than by the field's sources; the explanation says why it was made.

The Web address source downloads a data file from an address you paste in.

Use it for a file whose address you already have: a link from a colleague, a supplementary file of a paper, or a service Fit has no source of its own for. The address must begin with http:// or https://.

Which reader opens the file is decided by the name at the end of the address, and by what the server says the file is called when the address names nothing.

Limitations

An address that answers with a web page rather than a data file downloads that page, which then reads as no data points. The preview shows this before anything becomes a project.
A file on this computer needs no download: open it with File > Import Profile.

Published data (DOI)

This software's choice — A decision made by this software rather than by the field's sources; the explanation says why it was made.

The DOI source opens a Zenodo or figshare record by its DOI and lists the files it publishes.

Paste a DOI - 10.5281/zenodo.1234567, or the same thing as a doi.org link - and Fit asks the DOI system where it resolves to, then asks that repository which files the record holds. Choose one and it is previewed and imported like any other file.

Words rather than a DOI search Zenodo for records whose description matches them.

Limitations

Only Zenodo and figshare publish a file list Fit can read. A DOI that resolves anywhere else is refused, naming where it went, so that the file can be fetched with the Web address source instead.
A record may publish its data in a form no reader here understands - an archive, a spreadsheet, an instrument's own format. Those files are listed and greyed rather than hidden, so that what the record contains is still visible.