-
Created a tool that estimates price of laptops on noon store to help users descide which company they will buy from based on their budget.
-
Scraped over 863 laptops descriptions from noon store using python,beautifulsoup and selenium
-
Engineered features from the text of each laptop description to extract the features that affect on the price of laptops.
-
Optimized Linear, Random Forest, and XGboost Regressors using GridsearchCV to reach the best model.
-
Built a client wep app using streamlit
- language:python
- libraries:pandas, numpy, sklearn, matplotlib, seaborn, selenium, beautifulsoup, pickle,streamlit to install the libraries use this command:
pip install 'name of library'I scrapped over 863 laptops from noon website and extract from the description the following features:
- price
- Category
- Inch
- Core
- Processor
- Ram
- Memory
- Opysys
- after i scrapped the data i needed to clean it up,so i check for missing values and apply some technique to fill it.
- change the datatypes to make it valid for the model
- remove unusful characters from each feature
I looked at the distributions of the data and the value counts for the various categorical variables. Below are a few highlights:
First, I transformed the categorical variables into dummy variables. I also split the data into train and tests sets with a test size of 20%.
I tried three different models and evaluated them using Mean Squared Error.
I tried three different models:
- Multiple Linear Regression – Baseline for the model
- Random Forest Regressor – Because the ensamble model work effectivly and higher performance.
- XGboost Regressor – Again, with the sparsity associated with the data, I thought that this would be a good fit.
after experimant i found that the RandomForest had the less error in both training and testing so i fine tune them using grid search to have more accurate performance
In this step, I built a streamlit web app that was run on a local device, it requests list of values from user and returns an estimated price.
main page:
