-
Notifications
You must be signed in to change notification settings - Fork 20
Description
Is your feature request related to a problem? Please describe.
Currently, GoPredict focuses on predicting trip durations and displaying results through its pipeline and output files. However, there’s no dedicated user-facing section that allows users to view their previous trips and predictions.
This makes it difficult for users to analyze past results, compare predictions over time, or gain insights into their trip patterns.
Describe the solution you'd like
Add a Profile Section (either as a web UI component or dashboard module) where users can:
View a history of all past trips with details like:
Trip ID or timestamp
Source and destination
Actual vs. predicted duration
Weather or traffic conditions (if available)
Optionally filter trips by date, model used (e.g., XGB, RF), or performance metrics.
Export or download their trip history as a CSV or PDF.
Proposed Approach
Backend/Database
Create a simple database table or local storage mechanism to store trip data (trip metadata, model used, prediction results, etc.).
Automatically append trip details each time a new prediction is made.
Frontend/UI
Build a Profile page or dashboard (if integrated with the Netlify web app).
Display past trip records in a clean table or card layout.
Add filter and sort options (by date, model, RMSE, etc.).
Integration
Hook this feature into the main pipeline’s output generation step.
Whenever a prediction file is created, automatically log its summary into the Profile data.
Benefits
Improves user experience by allowing easy access to historical predictions.
Helps users understand model accuracy over time and patterns in trip durations.
Opens the door for analytics features, like weekly summaries or performance comparisons.
Additional Context
This feature aligns with GoPredict’s goal of being a comprehensive ML solution.
It can also serve as a foundation for future enhancements like:
Real-time prediction dashboards
User-specific analytics
A/B testing between different models
@harshitaphadtare Assign this to me