koaning
Series Generator
Feature Request: Line Drawing with Dual Y-Axis, time-range slider, and customizable dataset granularity
Overview: We propose a new feature for drawdata package that will allow users to draw lines directly within a Jupyter Notebook. The drawn lines will then be converted into dataframes, with additional functionality for customizing the visualization. The feature will include support for dual y-axes, a timeline slider for adjusting the x-axis, and an argument to set the granularity of the generated dataframe.
Key Features: Dual Y-Axis Support: The functionality will allow users to plot data with two y-axes, enabling the visualization of features with different scales on the same plot. The user will be able to select which feature should be plotted on each of the y-axes. Eg. Temperature and Radiation are highly correlated but have completely different ranges. Specification: Y-axis can support a total of 4 lines. Use can select the color these 4 lines.
Timeline Slider for X-Axis: A slider will be provided at the top of the plot to adjust the timeline of the x-axis. This feature will allow users to visualize how the dataset evolves over time, with the x-axis representing the timeline. The slider will control the visible range of the x-axis, offering a dynamic way to explore time-based data.
Granularity Argument/Dropdown An argument will be added to control the granularity of the dataset. This will allow users to specify the resolution or interval of the data points in the generated dataset. Eg. 5min/15min/Hour. Specification - Granularity will be equally spaced. A single dataframe cannot have multiple granularities.
Why This Feature? As a Data Scientist who frequently works with time-series data, I often begin my analysis by visualizing patterns through plots. However, I have found a gap in the tools available for this purpose. Currently, I have to rely on extracting data via APIs and other sources before I can begin exploring patterns.
What I am missing is a tool that allows me to immediately plot different features and generate the corresponding dataframe directly from the visualization. This feature would greatly enhance the workflow for data exploration and pattern recognition, allowing me to bypass the time-consuming data extraction process and move straight to the analysis.
Future Additions
- Bandwidths
- allow predefined curves like sine/tan etc.
- Allow more than 4 lines for Y-axis
