Interactive data exploration and forecasting tool
Role
Industry
Duration
Context
Sensitivity analysis (aka “what if” analysis) lets users interactively explore their data and see how different hypothetical scenarios might impact an outcome or variable. This is especially important for ensuring the fairness of a dataset.
Users score a trained model on a row, rows, or a whole dataset of potentially interesting simulated values and compare the outcome to the outcome of the original data. Users can filter, select and partition data; set new data values; selectively change data to simulate different circumstances or outcomes; compare model outcomes. This lets them see how changes cause the model predictions to change compared to the original data.
Capturing and communicating user research
I synthesized the findings from the user research into a series of personas. This is the primary persona; the data scientist. This lightweight persona summed things up nicely for this project. A better persona might be the more complex, enterprise-focused persona I use in participatory design sessions (on the right)
Usability testing, outcomes and future work
I created user journeys for the “happy path” for each persona’s primary task. Ideally, I would have included error cases also. I started exploring design ideas once there was a firm consensus around the requirements. I typically start with paper sketches and then move on to clickable prototypes. These illustrations aren’t meant to show a navigation flow. Rather, they are specific feature ideas.
