Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Excel 2.0 – Is there a better visual data model than a grid of cells?

subset.so

Post reply on HN
Print view View original

71–80 of 223 posts

Re: Excel 2.0 – Is there a better visual data model than a grid of cells?

#71

Post by rwhaling »

I've been thinking about this a ton lately - I think the holy grail is something that extends the spreadsheet UX with the capability to address much larger data sets, roughly equivalent in power to a Jupyter notebook.

This probably means constraining some of the totally free-form data entry and imposing some kind of discipline onto columns? As well as some way to intuitively group together groups of cell-level formulas into a logical "row", and describe row-level operations, with some kind of rough type checking or duck typing?

But you might want to still retain an ability to lay out the result of computations free-form for invoice and other reporting uses that Excel is still great for?

Better graphical visualization would be nice too and I think that's actually in reach - there are great open source options like vegalite and apache echarts that open up a lot of possibilities for new tools.

Re: Excel 2.0 – Is there a better visual data model than a grid of cells?

#72

Post by _old_dude_ »

gnramires wrote: post #67
AQXt wrote: post #64

Excel is great for simple use cases. For complex problems, you should look at Jupyter notebooks. https://jupyter.org/

I agree, I think it's a natural successor to a spreadsheet for something more complex. You can even have a section with tabular data (i.e. an embedded spreadsheet), which is what spreadsheets excel at, but it meshes very well with the classical programming paradigm.

Jupyter notebook is a regression from Excel, at least Excel computes the dependencies between the cells. With Jupyter, you have to be careful in the order you evaluate the cells.

And both falls on the floor the minute you want to use a spreadsheet/notebook as a library of functions, i.e it's not composable.

Re: Excel 2.0 – Is there a better visual data model than a grid of cells?

#73

Post by SebastianKra »

Maybe it's possible to find ways to better represent relational data?

Two interesting approaches:

- Ultrorg [1] attempts to represent relational databases in an excel-like format. You can build a query by editing the header of the table, and then edit the results in place.

- Tableau [2][3] is a visualisation construction tool, where users can describe a graph by assigning properties to rows or columns. The result can be a table of visualisations, for example a scatterplot matrix.

[1]: https://www.youtube.com/watch?v=tGzJ8F9rC1s [2]: https://www.tableau.com/products/desktop [3]: it's based on Polaris https://doi.org/10.1109/2945.981851

Re: Excel 2.0 – Is there a better visual data model than a grid of cells?

#74

Post by lhh »

I agree that spreadsheets are still extremely useful, and surprisingly difficult to improve upon. Maybe it's because spreadsheets combine a bunch of the primitives used to build many types of applications - tabular data, persistent data, operations on that data (CRUD, per the article), trivial to inspect the data, and can make various views on the data.

It seems to me that quite a few successful software companies boil down to a spreadsheet with enforced structure to the data and codified operations performed on the data specific to the problem domain. CRMs, accounting systems, project management systems, and ordering & inventory management systems are some examples.

Our product focused on financial projection & reporting (https://www.modeloptic.com/) is another example: Excel-like functionality at the core, and since we've constrained the domain and know how different pieces of data relate to each other, we can automate away a lot of the manual labor that'd be needed in completely free-form Excel.

Re: Excel 2.0 – Is there a better visual data model than a grid of cells?

#75

Post by Enginerrrd »

rwhaling wrote: post #71

I've been thinking about this a ton lately - I think the holy grail is something that extends the spreadsheet UX with the capability to address much larger data sets, roughly equivalent in power to a Jupyter notebook. This probably means constraining some of the totally free-form data entry and imposing some kind of discipline onto columns? As well as some way to intuitively group together groups of cell-level formul…

You know, despite being quite different than what you described, what comes close for me in concept is the Spyder IDE. They have pretty reasonable data visualization tools when you go to inspect variables. I find it really nice work in. It lets me do the real visualization and interplay between variables in code or in my head and just quickly inspect tables/vectors/rasters to sanity check as I work in real time.

What's missing is the ability to then manipulate those tables and have it backfeed into the code, but I think that honestly you could again get close to that with some basic right-click>generate new column> type in python expression > translate to input into interpreter type of workflow.

I also know there are some python based excel type programs.

Re: Excel 2.0 – Is there a better visual data model than a grid of cells?

#76

Post by Devasta »

I don't think you can improve Excel too much, but you could the ecosystem around it: Microsoft should integrate a package manager. Not something public, but something companies can privately maintain and publish packages to, that would contain common functions for their specific business.

If I could have a way to download a macro to a spreadsheet that would automate database connection for certain data sets, that would be huge, and since it'd be centralized corps could even have critical sheets functions managed by version control.

Mission critical excel spreadsheets run huge sections of large corps, we can discourage it or provide tools to better manage it.

Re: Excel 2.0 – Is there a better visual data model than a grid of cells?

#77

Post by stjohnswarts »

mromanuk wrote: post #51
dragonwriter wrote: post #38

For almost all tasks, yes. But it differs a lot by task. The power of the spreadsheet model is that it is minimally acceptable for a wide array of tasks, not that it is usually optimal.

I think that main strength is familiarity and low learning curve. You can easily build and MVP with a spreadsheet. Understand your data, relations, patterns. The trap occurs later when you need to scale it

I often use Excel to model data operations before I get to programming in pandas, so you're spot-on.

Re: Excel 2.0 – Is there a better visual data model than a grid of cells?

#78

Post by oneoff786 »

omarhaneef wrote: post #4

The key is that you can examine the state of most (not all) "intermediate" variables so not only do you know the inputs, the outputs, but also everything that happened in between in all these different ways. Your calendar app adds 2 days for tasks? You can just see the cell where it says 2 to make sure that variable is set right. Its amazing, and we are just catching up to it with "always on" variable inspectors in I…

You can, however, name cells and refer to them in formulas. Much nicer to have descriptive names instead of $J$7 for constants or inputs.

Re: Excel 2.0 – Is there a better visual data model than a grid of cells?

#79

Post by odipar »

Visual is all nice and all, but I really fancy the 'computational' model of spreadsheets which is very easy to learn and apply IMO.

My latest find is the CUE language, which I believe is the first 'typed' version of a spreadsheet (in disguise). CUE is like spreadsheets, but on steroids!

Here is an example: https://cuelang.org/play/?id=r4VXwbEG185#cue@export@yaml

Re: Excel 2.0 – Is there a better visual data model than a grid of cells?

#80

Post by willhslade »

Devasta wrote: post #76

I don't think you can improve Excel too much, but you could the ecosystem around it: Microsoft should integrate a package manager. Not something public, but something companies can privately maintain and publish packages to, that would contain common functions for their specific business. If I could have a way to download a macro to a spreadsheet that would automate database connection for certain data sets, that wou…

You should be able to do this with Git, Git-XL and a function to Export / Import BAS files from within Excel
Post reply on HN