gorkemgoknar/londonhousingprice

London Housing Price model trainer and inference api

Python

London House Price prediction Demo project

Note: this project is intended for pipeline generation for training and inference

##Building Docker images

To build Trainer

Rename Dockerfile_Trainer to Dockerfile
Run in interactive environment
docker run -it -v /yourworkdir:/tf/workdir -w /tf/workdir -p 8888:8888 IMAGEID bash

To build default DNN model just run

python train.py
If you need to fetch new data from GCP you need to provide credential file python train.py --credential_file credential_file.json

To build Inference API

Rename Dockerfile_Inference to Dockerfile
Run image which serves model
docker run -d -p 8080:8080 IMAGEID
use localhost:8080 to access serving api Api docs and sample can be found in localhost:8080/apidocs

Basic EDA

Seasonality

Area effect on price

Current Model performance