🏛️ Description

This project features 2 modes of deployments for the models built in the ML phase. The first is by api and the second is by using a web app.

API: https://immo-eliza-price-prediction-6csa.onrender.com

Web App: https://immo-eliza-deployment-webapp.streamlit.app/

Web App(independent of the API): https://immo-eliza-price-predict-rd.streamlit.app/

🏗️ Repo Structure

+---fast_api
|       main.py
|       
+---models
|       cat_boost.pkl
|       kitchen_ordinal.pkl
|       linear_regression.pkl
|       one_hot.pkl
|       random_forest.pkl
|       state_building_ordinal.pkl
|       
+---streamlit
|       slapp.py
|       using_api.py
|
|   .gitattributes
|   Dockerfile
|   README.md
|   requirements.txt

🛎️ How to use?

1. Open the web app link in a new tab and enter the values prompted to get a house prediction.
2. Use the API by following the specifications in the section below.

API Specifications

URL: immo-eliza-price-prediction-6csa.onrender.com

path: /get_price/

method: POST

Body:

Feature	Description	Accepted values
model	Name of the model.	random_forest, linear, cat_boost
property_type	Type of the property.	HOUSE, APARTMENT
subproperty_type	Sub-type of the property.	'APARTMENT', 'APARTMENT_BLOCK', 'BUNGALOW', 'CASTLE', 'CHALET', 'COUNTRY_COTTAGE', 'DUPLEX', 'EXCEPTIONAL_PROPERTY', 'FARMHOUSE', 'FLAT_STUDIO', 'GROUND_FLOOR', 'HOUSE', 'KOT', 'LOFT', 'MANOR_HOUSE', 'MANSION', 'MIXED_USE_BUILDING', 'OTHER_PROPERTY', 'PENTHOUSE', 'SERVICE_FLAT', 'TOWN_HOUSE', 'TRIPLEX', 'VILLA'
locality	Locality where the property is present.	'Aalst', 'Antwerp', 'Arlon', 'Ath', 'Bastogne', 'Brugge', 'Brussels', 'Charleroi', 'Dendermonde', 'Diksmuide', 'Dinant', 'Eeklo', 'Gent', 'Halle-Vilvoorde', 'Hasselt', 'Huy', 'Ieper', 'Kortrijk', 'Leuven', 'Liège', 'Maaseik', 'Marche-en-Famenne', 'Mechelen', 'Mons', 'Mouscron', 'Namur', 'Neufchâteau', 'Nivelles', 'Oostend', 'Oudenaarde', 'Philippeville', 'Roeselare', 'Sint-Niklaas', 'Soignies', 'Thuin', 'Tielt', 'Tongeren', 'Tournai', 'Turnhout', 'Verviers', 'Veurne', 'Virton', 'Waremme'
construction_year	The year in which the property was constructed.	Any integer year.
total_area_sqm	Total area of the property in square meters.	Any float value.
nbr_bedrooms	Numer of bedrooms.	Any integer value.
kitchen_type	Level of kitchen.	'NOT_INSTALLED','UNINSTALLED','INSTALLED','SEMI_EQUIPPED','HYPER_EQUIPPED'
fl_furnished	Is the property is furnished or not?	0/1
fl_open_fire	Does the property have a fireplace?	0/1
terrace_sqm	Terrace area of the property in sqm.	Any float value
garden_sqm	Garden area of the property in sqm.	Any float value
fl_swimming_pool	Does the property have a swimming pool?	0/1
fl_floodzone	Is the property in a flood zone?	0/1
state_of_building	State of the building.	'TO_RENOVATE','TO_RESTORE','TO_BE_DONE_UP','GOOD','JUST_RENOVATED','AS_NEW'
primary_energy_consumption_sqm	Energy consumption per square meter.	Any float value
heating_type	Heating type used by the property	'CARBON', 'ELECTRIC', 'FUELOIL', 'GAS', 'PELLET', 'SOLAR', 'WOOD'
fl_double_glazing	Does the property have double glazing windows?	0/1
cadastral_income	Cadastral income of the property.	Any float value

NOTE: All the accepted values noted above are case sensitive

Sample request

{
    "model":"cat_boost",
    "property_type":"HOUSE",
    "subproperty_type":"HOUSE",
    "locality":"Gent",
    "construction_year":2018,
    "total_area_sqm":70.4,
    "nbr_bedrooms":2,
    "kitchen_type":"SEMI_EQUIPPED",
    "fl_furnished":1,
    "fl_open_fire":0,
    "terrace_sqm":5,
    "garden_sqm":15,
    "fl_swimming_pool":1,
    "fl_floodzone":0,
    "state_of_building":"TO_RENOVATE",
    "primary_energy_consumption_sqm":100,
    "heating_type":"GAS",
    "fl_double_glazing":1,
    "cadastral_income":1200
}

Sample Response

{
    "price": 340896.68
}

Enhancements

1. Add validation to the API development.
2. Allow missing values in the API.

⏱️ Timeline

This project took five days for completion.

📌 Personal Situation

This project was done as part of the AI Boocamp at BeCode.org.

Connect with me on LinkedIn.