peng22/house_pricing_predictor

This is a project for machine learning model to predict house prices

Project Scope

This project is for predicting the house prices using housing_price_dataset.csv

https://www.kaggle.com/datasets/muhammadbinimran/housing-price-prediction-data

We made the data preparation then we used diffrent models and then chose the best model. we used the r2_score and mean square error to compare the models.

Using pickle we were able to export the chosen model to be used.

Probem Description

We have some info about the house like the area, neighbourhood, number of bedrooms and number of bathrooms using these data we try to predict the house price.

we tried to get the relation between these information and the price.

Then we did some processing to get the best performance.

Deploying the model locally

We created a Dockerfile to create an image.
firstly we can build the model using:
docker build . -t house_predictor then we can run the image using: \ docker run -it -p 9696:9696 house_predictor

then we can run the predict_local.py after installing requests in our environment as the url will be: url = 'http://localhost:9696/predict_house_price' and run
python predict_local.py

Using the deployed model

to use the deployed model the url will be:
url = 'http://peng24.pythonanywhere.com/predict_house_price'
then run
python predict.py

predicted dictionary shape

Make sure the predicted dictionary to be in this form:
bathrooms from 1 to 3
bedrooms from 2 to 5

client = {"squarefeet":2126, "yearbuilt":1969, "neighborhood":"Rural", "bathrooms":3, "bedrooms":4, }