Tempora: an Accessible Time Series Analysis App
Forecasting is a fundamental part of business, yet it’s often overlooked. Many organisations rely on basic tools (think Excel’s FORECAST.ETS function) even when their data could support more informed, forward-looking decisions.
To address this, I built a simple, accessible time series analysis and forecasting application called Tempora in Streamlit. The goal was to create a tool where users can quickly explore and forecast univariate time series data without writing code or setting up complex environments. Rather than a one-off solution for a specific problem, I wanted to develop a general tool that could be used across different sectors, saving time and avoiding repeated processes.
Time series forecasting can feel like an art. Trends, seasonality, and other components make it more complex than a straight-line projection. Many people shy away from it because it’s not obvious how to do it well. Yet, when done properly, it can have real business impact—helping with budgeting, planning, and operational decisions. My aim was to make forecasting accessible: clear, intuitive, and free from unnecessary complexity.
The app guides users naturally from uploading data to exploring it, transforming it, and generating forecasts. Users can visually inspect data, assess stationarity, run models like ARIMA and Prophet, and produce forecasts that are interpretable rather than black boxes. Tooltips and parameter explanations help users understand each step along the way.
This project also ties into my broader work in Data Archive, where theoretical topics such as stationarity and autocorrelation are documented. Concepts are hyperlinked within the app so users can quickly access deeper explanations if they want to explore further.
User experience has been a major focus in this version. The app is designed to provide real utility while bridging the gap between theory and practical application. It’s general enough to be useful for anyone working with time-dependent data, regardless of sector. Future updates may expand functionality based on user feedback, but for now, simplicity and accessibility are the priority.
The main goal of this post is to invite users to try the app and provide feedback. If you work with time series data and want a tool that makes forecasting approachable, I encourage you to explore it and share your thoughts.
Leave a comment