productivity.py
Generates and displays the user's productivity in the latest recording.
Usage:
1 Introduction
The purpose of this document is to define the requirements and design of the Productivity measurement system in the OpenAdapt process automation library.
2 Background
Motivation:
Useful for businesses to evaluate the performance of their employees through their productivity
Convenient for us to measure this productivity since we are recording the user’s actions
Businesses would be more inclined to use OpenAdapt if they knew how it helped with their productivity
3 Goals
We want to be able to provide businesses with metrics that show how productive a user is in a recording. These metrics include but are not limited to:
Number and length of repetitive tasks
Total/average time spent on repetitive tasks
Number of mouse clicks
Number of key presses
Number of long pauses
Number of window/tab changes
Total time spent on each window/tab
Number of errors (at the moment we don’t have a good way to do this)
4 Design
4.1 System Overview
4.2 Task identification
Recursively use the longest repeated non-overlapping substring algorithm to find repeating lists of ActionEvents until the final list contains a single task and no repetition
O(n^2), room for optimization
4.3 Window event data
Show one screenshot for each time the user switches tabs/windows, accompanied by data like # of clicks, # keypresses, and time spent on the tab
4.4 Visualization
Everything described above is visualized as an HTML page using bokeh, similarly to the original visualize.py
At the top of the HTML page there is data about the entire recording, including things like # of tasks completed, total/average time spent on repetitive tasks, # keypresses/clicks, # pauses, # tab changes
5 Analysis
5.1 Performance
Plots and discussion of relevant metrics (e.g. accuracy, throughput)
6 Future Work
Refactor to make a clean, sleek webpage rather than just visualize HTML
Ultimately, it would be nice to be able to compare the productivity of a user with that of a ReplayStrategy - time spent on completing the tasks in a recording and error rate would be a useful comparison
Once we have a vision model that is suitable for GUI images, we can use that model to identify the task identified as the longest repeating task in the recording
Ideally, we would get feedback from businesses about what information suits their needs
Last updated
