I produce data in Linux and use Grace in order to visualize data. However, Grace just works in 2D. Now, I have problems with Grace. I am looking for better tool.
I heard Gnuplot is a perfect tool in Linux but I have problem in its running.
I would not call gnuplot an ideal tool. It is free, and easy to use if you need to make a line-graph quickly, but for more than that it is not very useful.
VisIt and Paraview will run on a Linux machine, can plot 3-D data, and both are freeware (see links attached to this reply). Alternatively, you can plot using python, but that will require some basic programming skills, because you have to write your own routines. Of course, there are also plenty of commercial visualization tools, but those tend to be expensive
Take a look on the matplotlib for pyhton. If you do not mind setting up a python environment, of if you already have one, the matplotlib could be a very good choice:
http://matplotlib.org/
It gives you some neat little features, just like Matlab, but its free. The overhead to get python running will can require some time though...
The ROOT - Data Analysis Framework is an excellent tool to visualize data. The framework is pubished by CERN and is used to visualize all kind of data. Is supports 2D / 3D visualization. It offers a huge amount of graph fitting and error estiamtion techniques. And (for you the most important) it is designed for Linux systems!
Here is a Link to he page: http://root.cern.ch/drupal/
I myself do most of the work with mentioned QtiPlot (http://www.qtiplot.com/) which is, in design and functionality, qutite similiar to Origin. You may even have a look at SciDavis, a QtiPlot fork.
I would argue Gnuplot is powerful tool for scientific plots. However, the learning curve to produce complicated, publication quality drawings is quite steep. It is not just the simple line graph it is capable of but there is more. See the demo website of the current version gnuplot 5.
Gnuplot has many possibilities. You can use Linux bash pipe to provide the gnuplot script with data from other programs. There are even gnuplot bindings for python.
If you're still searching for the tool, I'd suggest BioVinci. You can find both 2D/3D graphs and statistical analyses in there. I've tried its 3D scatter plot, violin plot, and heatmap - all are interactive and easy to customize. Its PCA is pretty cool, too.
I would like to suggest MATLAB is all in all for every OS. See e.g., https://au.mathworks.com/help/matlab/matlab_env/start-matlab-on-linux-platforms.html for Linux.