Hi @schlerp , I happened to come across your medium article and had a similar idea parked for a while now: to profile existing data processing pipelines/scripts/notebooks to build a representation of the data flow. I believe this could be a future feature to this module, so this is very exciting to me. I work in Data Science and would find this extremely useful.

I'd like to contribute to adding new features on the pandas front, since I do not know much on the UI side of things. A few regular operations I use are listed below:

1. Basic feature/column computations
2. Aggregation using groupby
3. Resampling time indices
4. Replacing and imputing values
5. Pivot, melt and stacking

Before I try to modify anything, I'd like to know if there's a tracker of sorts for current development activity and upcoming features already being worked on.

Hi, I tried to install and run flowpy with the help of the instructions in the README.md. But at least on the machines (Windows 7 and MacOS) which I tried to run flowpy the yarn-program was always missing. Can you add instructions on how to install the missing program? It seems that it does not get installed by the python modules in the requirements.txt. For someone who does have it already installed, it could deter them from trying out flowpy.

Hi @schlerp, I have been thinking it would be useful provide input fields with drop downs to the user. As a reference, I use bamboolib for personal projects and they do this very well. Link: https://docs.bamboolib.8080labs.com/documentation/getting-started

Started this issue to discuss on a design for implementation since the same design would be usable for the majority of the app, ranging from column name selection all the way to hyper parameters to aggregation functions like datetime format/resampling frequency etc.

Boolean parameters will still be kept as check boxes, as has been implemented for index in csv output block.

This feature would allow adding conditions on the DataFrame and then assigning value for a new column

Example

A > 5 & B < 10; Create Flag as 1, else Flag as 0

This could then be used to create masks on the data to be used by the already present Filter block. (for future enhancements)

This feature would emulate the typical pandas groupby operations and provide calculating metrics like mean, count, min max etc

Example

Groupby key - Multiselect for columns Metric - mean, count, std, min, max etc.

Inclusion of a custom groupby metric (as a function) can be thought of for future enhancements

This feature would involve creating a new column from existing ones. Support for formulas like the excel formula bar would be the expected functionality.

Example

A**2 A + B / C A == C A = 0

Target column name would also need to be provided here, which can be a new column or an existing one.

I would like to start contributing from this feature, will generate a pull request once complete. Any thoughts?