/predicting_house_prices

This repo contains a Jupyter Notebook + API to develop a ML suite to predict property prices

Primary LanguageJupyter Notebook

Predicting house prices

This repo gives a comprehensive guide to predict house prices

Features

With this package you will be able to:

  • Using Property-Friends-basic-model.ipynb, analyse train and test set, data explore, data pre-process, grid search explore a Gradient Boost Regressor, train a Gradient Boost Regressor, load and evaluate the model in the test set

  • Run a API responsible to run the trained model and evaluate it on the test set using a docker

Requisites

All the code is done in Python, txt and Markdown. To run the pipeline (Jupyter Notebook) and model deploy it on a Docker (notice that they are on requirements file in this case), you will need the following Python libraries:

  • pandas
  • numpy
  • matplotlib
  • scikit-learn
  • seaborn
  • pickle5

Usage

In this repo you will find the 3 main components:

  • Property-Friends-basic-model.ipynb: which can be used as a notebook to get the model to be deployed. Use this as a typical notebook
  • model_deploy.py: contains the instructions necessary to run the model (model/pima.pickle.dat)
  • Dockerfile: contains docker instructions to build the image

To build the image, go to the main project directory and use: '''$ docker build -t predicting_house_prices:latest .'''

To run the image: '''docker run predicting_house_prices:latest'''

To do

  • Expand the range and hyperparameters in the GridSearch
  • Include KFold/CV splitter
  • Improve the logs related to the whole pipeline
  • Include the security system for the API