# Scatterplot with data from CSV file and trendline

I have created a scatterplot with code from this previous asked question.

\documentclass[varwidth=true, border=2pt]{standalone}
\usepackage[utf8]{inputenc} % this is needed for umlauts
\usepackage[ngerman]{babel} % this is needed for umlauts
\usepackage[T1]{fontenc}    % this is needed for correct output of umlauts in pdf
\usepackage[margin=2.5cm]{geometry} %layout
\usepackage{pgfplots}

\begin{filecontents}{table3.csv}
column 1    column 2
1966    37.51817228
1960    40.56693583
1961    40.71972964
1962    40.97560208
1964    41.11687187
1963    41.25082828
1965    46.02625404
1960    46.22815872
1967    46.67800113
1961    48.39523271
\end{filecontents}

\begin{document}

\begin{tikzpicture}
\begin{axis}[
axis x line=middle,
axis y line=middle,
enlarge y limits=true,
%xmin=0, xmax=2150,
%ymin=0, ymax=600,
width=15cm, height=8cm,     % size of the image
grid = major,
grid style={dashed, gray!30},
ylabel=steps,
xlabel=$n$,
legend style={at={(0.1,-0.1)}, anchor=north}
]
\addplot[scatter,only marks] table [x=column 1, y=column 2, col sep=comma] {table3.csv};

%the code below is added via @Peter's comment.
\addplot[only marks] table [col sep = comma,y={create col/linear regression={y=column 2}}]{table3.csv};

\end{axis}
\end{tikzpicture}

\end{document}


The scatter plot ends up great but I want to be able to add a trendline. All the examples of trendlines I've seen have been calculated with data inputted directly into the .TeX file and not from a .csv file.

Is it possible to do this?

My other thought was to go on on excel, calculate the trendline, and then overlap the line onto the graph. I'd really rather be able to do it in a more direct fashion though as my document has many graphs.

edit: Jake gave me great guidance on how to do this with data directly inputted to the TeX file but I am having trouble with parsing directly from the .csv file. I have added to my code and posted the error message I get in the console.

With the added code I get the error message.

./linearreg.tex:30: Package PGF Math Error: Could not parse input '' as a float
ing point number, sorry. The unreadable part was near ''..


Line 30 in my document is the added line with the linear regression equation.

final edit: I figured it out. This error was caused because my file had columns with blank data. I had to delete this blank data to calculate the linear regression line.

Here is the final mwe of my code:

\documentclass{article}
\usepackage{tikz}
\usepackage{pgfplots}
\usepackage{pgfplotstable}

\begin{document}

\begin{tikzpicture}
\begin{axis}[
xlabel=Weight (kg), % label x axis
ylabel=Height (cm), % label y axis
axis lines=left, %set the position of the axes
clip=false
]

\end{axis}

\end{tikzpicture}
\end{document}

• Can you include a few lines from that table3.csv file? You can look at the pgfplots manual for linear regression (another name for trendline). Oct 5 '14 at 9:19
• @Cain: You use col sep=comma, but your data file isn't comma-delimited. Either add commas between the columns of your data file, or remove the col sep=comma and choose column names that don't have spaces in them.
– Jake
Oct 5 '14 at 15:11
• @Jake figured it out! It was due some blank entries in my file. Thank you for the guidance.
– Cain
Oct 5 '14 at 15:28
• Maybe you should remove the code sample, since it apparently does not represent the structure of your actual data file. As it stands, this is potentially confusing for others who come across the error message you posted. The final code you posted is meaningless without a sample of the data file.
– Jake
Oct 5 '14 at 15:32

To get a linear regression line for data from a data file, use

\addplot [no markers] table [y={create col/linear regression={y=<column name>}}] {<file name>};


\documentclass[border=5mm]{standalone}
\usepackage{pgfplots, pgfplotstable}
\usepackage{filecontents}

\begin{filecontents}{table.dat}
x y
0 1
100 3
150 2
200 5
300 6
\end{filecontents}

\begin{document}

\begin{tikzpicture}
\begin{axis}[
axis x line=middle,
axis y line=middle,
enlarge y limits=true,
width=15cm, height=8cm,     % size of the image
grid = major,
grid style={dashed, gray!30},
ylabel=steps,
xlabel=$n$,
legend style={at={(0.1,-0.1)}, anchor=north}
]
\addplot [no markers, thick, red] table [y={create col/linear regression={y=y}}] {table.dat};
\end{axis}
\end{tikzpicture}

\end{document}

• Thanks for this. It works when I input the data natively in the TeX file but for some reason it has difficulty parsing directly from from the .csv file. Which I need to do because I'm dealing with a lot of data. Here's the error I get: Package PGF Math Error: Could not parse input '' asa floating point number, sorry. The unreadable part was near ''.. I will add my edited code and error message above.
– Cain
Oct 5 '14 at 14:45
• @Cain: As percusse said in his comment, you should include a few lines of the data file in your question.
– Jake
Oct 5 '14 at 14:52
• added. Those are just the first few lines. There are ~1800 hundred data points.
– Cain
Oct 5 '14 at 15:02