Team Innovators - 159

Architecture of the Application
UI of the Application

Inspiration

The inspiration to design and build this project came from the problems that people face while searching an ideal house/property but can't find it because of pricing and capital issues. Definitely buying an expensive house is not everyone's cup of tea. So, it is cumbersome for some people to buy houses within their budget. And so we got an inspiration to build an interface for potential buyers to get in touch with the house of their desire with a satisfactory price that suits their need and save their time by comparing prices.

What it does

This is an application in which we can just feed in the location, area, availability, square footage, BHK and a bathroom and we can know the price of the house we need.

How we built it

We set up LinuxOne cloud on virtual machine and installed jupyter lab on it.
Then we chose the dataset from kaggle.
We cleaned the dataset as per our need.
Then in the jupyter notebook we wrote the code for data preprocessing.
Then we split the train and test data.
We applied the Linear Regression model with the accuracy of 82%.
Next model we applied is lasso with the accuracy of 81.2%.
The next model we used is Ridge with the accuracy of 82.3% and imported pickle.
We dumped our ridge model with the name RidgeModel.pkl.
And then we imported cleaned_data.csv and the RidgeModel.pkl inside our IDE.
And then we made a simple skeletal structure of flask.
Here the libraries we used are flask, Scikit-learn, Pandas, Pickle-mixin. So, in our terminal first we imported all the libraries.
We created a template called as index.html where our model page is rendered out.
We used bootstrap for the general design of our application.
We integrate our ML model and HTML file by sending the value of the object using render_template. ## How it works 1) Collecting Data: First step was to collect data. We collected data from different sources and merged them together to form our training data set.

2) Then we trained the model using machine learning algorithm which in this case is multiple linear regression.

3) Based on the generated graphs, we predicted the cost of the house

Challenges we ran into

The first challenge we ran into was setting up linuxone cloud on our virtual machines. After making the account and getting the private key we went onto download the putty but when we actually downloaded the file for installation a lot of it was not compatible with the machines of most of our team members. Once we got the ideal software we went on with further setup and the key was inactive. After solving that again we landed up in errors but somehow we successfully managed to setup our machines. After setting up the machines an other issue was downloading docker but with errors and time going from our hands our confidence didn't go and we kept trying and the setup was done. Now we needed something that is called as a project sadly we had to cancel two potential projects because of lack of time. We tried our best and made this project called as House price prediction application and even making it was a task as errors never left our way. While being with each other throughout we did it finally. But also if wee say in general our model was in pickle so it was very difficult for us to deploy our project on IBM system because the tutorial guide only had tutorial for pytorch and tensorflow conversions to ONNX. No such tutorial was given for pickle. So it was a drawback and we faced a lot of issues because of that.

Accomplishments that we're proud of

We are proud that we were able to setup our machines.
We are proud we were able to work together well.
We are proud that together we build our project.
We are proud that we learnt some new technologies during the course of designing and building our project.