HN Explorer

Show HN: Marple – Interactive time series visualization for engineers (marpledata.com)

42 points by NeroVanbierv ↗ HN
Hi HN, excited to share the project I have been working on for the past two years. Marple was created by engineers for engineers. Marple helps them organize and analyze their sensor data in a more user-friendly way.

We’ve all been there: measurement data from an experiment or test needs to be analyzed but the only thing you have is a few data files flying around. You do not have a clue how to open or read the data. So what you do next is open your Python/Matlab/Jupyter Notebook/… and start coding in order to make some sense of the data. We didn’t like that process, so we started Marple to solve this issue.

Engineers tend to log data at frequencies from 1Hz to 10kHz and usually log hundreds of sensors at the same time. Data sets usually contain millions of data points. In order to make our web-based data visualization responsive, we had some technical challenges to tackle. We developed our own visualization engine based on PostgreSQL that is able to visualize millions of data points pretty much instantly. This allows us to create an interactive visualization environment which is perfect for data exploration, even for large data sets!

This is the second time we show Marple to HackerNews, but since then we made some big steps. We made a pivot to a cloud product and now offer a free version of Marple. Feel free to head over to our website and give it a go. Let us know what you think of it!

14 comments

[ 3.5 ms ] story [ 44.7 ms ] thread
xvilka
I think, name isn't the best one - I initially read it as "Maple", especially in a context of scientific computing and visualization.
NeroVanbierv
We've indeed had other people confusing us with Maple before! Luckily Maple is not so often used in an engineering data context.
hamter
This looks pretty useful and I will definitely be trying it out. I see you can upload CSV files for visualizing. Is there a way to send data as it's generated?

Another question would it be possible for Marple to use the data locally (or from say my own s3 bucket or database) instead of having to upload it to your servers?

NeroVanbierv
Cool, thanks!

Is there a way to send data as it's generated? Yes! We have a tiny python SDK that can take care of that. Read more here: https://www.marpledata.com/docs/by-category/developers

[...] use data locally instead of having to upload it to your servers? No, not yet. We are however prototyping an influxdb integration. Would that cover your use case?

hamter
I haven't used influxdb before but if the data resides on hosts I own I think that would be acceptable. Thanks for your answers!
alphakappa
Unrelated to the product itself, please make pricing/licensing clear upfront. “Try for free” leads to a page that says “signing up is free”. Great, but what comes after that? Please don’t leave your users guessing.
MBaert
You make a valid point, there is a completely free version actually, but you can also try the paid version 14-days for free. After signing-in you can choose the free plan or the trial, but I get your point, you might feel like you are running in some sort of trap. Maybe we could make a link to the pricing page there to make it more clear.

PS: I also work at Marple

huevosabio
Looks really cool! I work with phone IMU and this looks appealing.

We created our own in house infra to manage and visualize the data, but will try to check this out.

MBaert
Nice! If you have accel, gyro and maybe even GPS data then this might a really good use case!
mojomark
Looks like plugins available for Matlab/Simulink from the graphic on your intro page. Didn't see a quick list of supported interfaces - do you support: Dymola FMI? Modbus?
MBaert
We support .mat files, we do have a small example how to integrate with Matlab, but we do not have a proper SDK or Simulink integration (yet).

I'm not too familiar with Dymola FMI or Modbus. Good ideas for the future :)

d4rkp4ttern
What specifically makes it only (or mainly) relevant to “engineers with sensor data”? It seems like it can be broadly useful for any (numeric) time series data ?