https://github.com/sureshbeekhani/gemstone-price-prediction

The Gemstone Price Prediction project aims to predict gemstone prices using a machine learning approach. The project involves various steps such as data ingestion, transformation, model training, and deploying the model through a Flask web application.
https://github.com/sureshbeekhani/gemstone-price-prediction

data-science machine-learning predictive-modeling regression-analysis

Last synced: over 1 year ago
JSON representation

The Gemstone Price Prediction project aims to predict gemstone prices using a machine learning approach. The project involves various steps such as data ingestion, transformation, model training, and deploying the model through a Flask web application.

• Host: GitHub
• URL: https://github.com/sureshbeekhani/gemstone-price-prediction
• Owner: SURESHBEEKHANI
• License: other
• Created: 2024-10-20T05:03:06.000Z (almost 2 years ago)
• Default Branch: main
• Last Pushed: 2024-11-14T05:35:14.000Z (over 1 year ago)
• Last Synced: 2025-02-02T08:13:21.849Z (over 1 year ago)
• Topics: data-science, machine-learning, predictive-modeling, regression-analysis
• Language: Jupyter Notebook
• Homepage: https://sureshbeekhani-gemstone-price-prediction.hf.space/
• Size: 15.3 MB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# Gemstone Price Prediction - Suresh beekhani

### Introduction About the Data :

**The dataset** The goal is to predict `price` of given diamond (Regression Analysis).

There are 10 independent variables (including `id`):

* `id` : unique identifier of each diamond

* `carat` : Carat (ct.) refers to the unique unit of weight measurement used exclusively to weigh gemstones and diamonds.

* `cut` : Quality of Diamond Cut

* `color` : Color of Diamond

* `clarity` : Diamond clarity is a measure of the purity and rarity of the stone, graded by the visibility of these characteristics under 10-power magnification.

* `depth` : The depth of diamond is its height (in millimeters) measured from the culet (bottom tip) to the table (flat, top surface)

* `table` : A diamond's table is the facet which can be seen when the stone is viewed face up.

* `x` : Diamond X dimension

* `y` : Diamond Y dimension

* `x` : Diamond Z dimension

Target variable:

* `price`: Price of the given Diamond.

Dataset Source Link :

[https://www.kaggle.com/competitions/playground-series-s3e8/data?select=train.csv](https://www.kaggle.com/competitions/playground-series-s3e8/data?select=train.csv)

### It is observed that the categorical variables 'cut', 'color' and 'clarity' are ordinal in nature

# Deployment Link :

 link : [https://sureshbeekhani-gemstone-price-prediction.hf.space/](https://sureshbeekhani-gemstone-price-prediction.hf.space/)

# Screenshot of UI

# YouTube Video Link

Link for YouTube Video : Click the below thumbnail to open 

![API Prediction](./Screenshots/Prediction.jpg)

# Approach for the project 

1. Data Ingestion : 

    * In Data Ingestion phase the data is first read as csv. 

    * Then the data is split into training and testing and saved as csv file.

2. Data Transformation : 

    * In this phase a ColumnTransformer Pipeline is created.

    * for Numeric Variables first SimpleImputer is applied with strategy median , then Standard Scaling is performed on numeric data.

    * for Categorical Variables SimpleImputer is applied with most frequent strategy, then ordinal encoding performed , after this data is scaled with Standard Scaler.

    * This preprocessor is saved as pickle file.

3. Model Training : 

    * In this phase base model is tested . The best model found was catboost regressor.

    * After this hyperparameter tuning is performed on catboost and knn model.

    * A final VotingRegressor is created which will combine prediction of catboost, xgboost and knn models.

    * This model is saved as pickle file.

4. Prediction Pipeline : 

    * This pipeline converts given data into dataframe and has various functions to load pickle files and predict the final results in python.

5. Flask App creation : 

    * Flask app is created with User Interface to predict the gemstone prices inside a Web Application.

# Exploratory Data Analysis Notebook

Link : [EDA Notebook](./notebook/1_EDA_Gemstone_price.ipynb)

# Model Training Approach Notebook

Link : [Model Training Notebook](./notebook/2_Model_Training_Gemstone.ipynb)

# Model Interpretation with LIME 

Link : [LIME Interpretation](./notebook/3_Explainability_with_LIME.ipynb)