Seer: Forecasting Microservice

What is this?

Seer is a service that can do real-time forecasting on time series data. You can run it using one of the provided docker images, then talk to it using a gRPC client.

It allows you to simply stream in your data as you receive it, then generate up to date forecasts.

Why did I build it?

Seer is the open-source release of a product I built to provide time series forecasting as a service. The original product was a multi-tenant cloud API, but this release is a single-tenant version of that product intended for internal corporate or personal use.

How do I use it?

To run a copy of the server locally, use the latest docker image:

docker run -d -p 8080:8080 cshenton/seer

Then use one of the available clients to stream in data and start forecasting:

• go, go get github.com/cshenton/seer-golang/...
• python pip install seer

Roadmap

• Speed and accuracy benchmarks
• Demo

Other

Generating server snippets

protoc -I seer/ seer/seer.proto --go_out=plugins=grpc:seer